SELIM03.pdf


45

Ricardo Mairal & Javier Martín, Selim  3 (1993): 43-56

UNDERLYING PREDICATIONS AND LATIN-OLD ENGLISH

TRANSLATION: TWO PREDICATES UNDER SCRUTINY1

0. INTRODUCTION

The aim of this paper is to study the translation into Old English of two 

Latin verbal predicates: deliberare/Íencan and amitto/gewitten2. For this 

purpose, we will resort to one of the most elaborate theoretical linguistic 

models, viz. Functional Grammar (henceforth FG)3.

One of the key notions of FG that plays a fundamental role in our transla -

tion study is that of underlying predication. Dik (1979) claims that predica-

tions are characterized in such a way that makes them the best representation 

1 This article is a revised version of the paper “Underlying predications and translation: 
on the Old English version of a Latin text”, which we presented in the 7th
International Congress of English Historical Linguistics, held in Valencia (September 
1992).

2 From The Old English Apollonius of Tyre (in Goolden 1958:6 -7).
3 The following abbreviations are used in this paper: 
(i) predicative categories: V (verb);
(ii) syntactic functions: S (subject) and O (object);
(iii) semantic functions: Ag (agent) Loc (Locative) and Exp (expiriencer);
(iv) operators: p3 (degree of certainty, an operator that modifies the propositional 

content of the clause);
(iv) phrasal categories: NP (noun phrase) and PP (prepositional phrase);
(v) the hierarchical structure of the clause: E1(variable designating state of affairs), ILL 

(illocutionary force) and PRED (predication);
(vi) translation terms: L1 (source language) and L2 (target language).



46

of sentences. Then, on the basis of this idea, a translation model (Dik 1979 

and 1986; Van der Korst, 1990) is formulated by means of describing the 

procedure for arriving at the mechanical translation of underlying predica-

tions of one language into underlying predications of another language.

In this paper we will investigate, on the basis of the information provided 

by the underlying predications, to what extent the underlying predication of 

the source language (Latin) can be related to an equivalent underlying predi-

cation of the target language (Old English).

1. UNDERLYING PREDICATION

Dik (1989) proposes to treat clausal constituents under the form of an ab-

stract underlying clause structure, which will be mapped onto the actual lin -

guistic expresions by a number of expresion rules, which provide these con-

stituents with the order, form and intonation pattern (Dik l989:45). Then the 

analysis of the clause entails the distinction of several layers or levels of 

formal and semantic organization. These layers would be illustrated in the 

following chart:

(1)

CLAUSE “speech act”

PROPOSITION “posible fact”

PREDICATION “state of affairs”

PREDICATE “property/relation”

This type of analysis has been called the layered hypothesis. It was origi-

nally proposed by Hengeveld (1987,1988 and1989), Dik (1989) and Dik and 

Hengeveld (1990). These linguists propound to represent utterances by 

means of a multilayered clause model. Dik and Hengeveld (1990:2-3) support

the view that two main levels should be distinguished in an analysis of the



47

clause: the representational and the interpersonal level, each being modified

by a number of operators and satellites. By the higher level or interpersonal

they mean that concerned with those linguistic means which are used by the 

speaker to evoke a certain comunicative effect in the addressee, whereas the 

lower level or representational is concerned with those linguistic means 

which are used by the speaker to provide the addressee with a description

of a state of affairs.

The representational level is  built on the basis of (i) a predicate frame 

(Level 1), which specifies a predicate plus the number of arguments required, 

and (ii) a predication (Level 2), which locates the State of Affairs in terms of 

its time ocurrence, frequency of ocurrence and actuality of occurrence. The 

illocutionary level is constructed on the basis of (i) an illocutionary frame 

(Level 4), which signals the value of a sentence as product of a speech act, 

and (ii) the propositional level (Level 3) which represents the sentence as a

possible fact, something which can be verified or denied.

Then, underlying predications will have the form as drawn in (2):

(2)

LAYERS
Clause  (E1: [ ILL (S) (A) (X1: etc. (X1))] (E1))

Proposition (X1: [ e1: etc. (e 1))] (X1))

Predication (e1: [Pred. (x1)] (e1)

Term (x1: Pred (x1))

The structure underlying linguistic expresions consists of four hierar-

chichally ordered layers1, each one designating different entities as proposed 

in Lyons (1977).

1 Dik and Hengeveld distinguish a number of operators and satellites in each layer (for 
further reference of these operators and satellites we refer the reader to the works 
mentioned above).



48

We will use this hierarchical model not only for the descrip tion of sen-

tences but also for a description of clausal complements. In some cases, we 

will see how the different clausal complements signal different types of layers

namely predications, propositions and utterances (Dik and Hengeveld 1990).

In sum, the information specified in the underlying predications does not 

only contain syntactic and semantic features but also pragmatic and proposi-

tional. These latter two will play a conspicuous role in translation studies 

since they provide more tools to achieve a higher degree of accuracy in trans-

lation.

2. UNDERLYING PREDICATION AND TRANSLATION

As pointed out above, underlying predications will be the starting point 

for our translation research. In line with Dik (l979 and 1986) and van der Ko rst

(l990:290), a translation procedure ammounts to three phases:

(i) reconstruction of the underlying predication of a sentence,

(ii) substitution of the underlying predication according to the equiva-

lences and

(iii) application of the expression rules of the target language.

This paper will concentrate on the first two leaving for the moment those 

issues dealing with expression rules.

Then, our first task will consist of reconstructing the underlying predica-

tion for both the source language (Latin) and the target language (Old En -

glish). In doing so, we will follow the main postulates of FG, succintly pre -

sented in the previous section.

The second phase of our work -perhaps the most interesting- will be con-

cerned with the (non-)equivalences of the expressions under scrutiny. Once 

we have the underlying predication for a predicate of L1 and a predicate of 



49

L2, we will see to what extent these are equivalent and therefore can be sub-

stituted.

This substitution of predicates involves a number of factors:

-quantitative and qualitative valency of the two predicates;

- syntactic and sematic functions;

- selection restrictions and

- type and nature of clausal complements.

Summarizing, we have presented so far the fundamental ideas that guide 

our research in this paper. In the third section, these ideas will be put to the 

test of translating two underlying predications of a Latin text into the corre -

sponding Old English predications.

3. TWO PREDICATES UNDER SCRUTINY

3.1. deliberare/Íencan

Let us detain now to determine the semantic weight of the first two lex-

emes that we have chosen, given the limitations of this paper, to carry out our 

analysis. The semantic weight will serve to evaluate their differences and, 

ultimately, “to make a choice”.

As a preliminary step, the determination of this semantic weight is based 

on the analysis of both the paradigmatic and syntagmatic axes since we be-

lieve that the intersection of both constitutes the meaning of a word. Follow-

ing the methodological premises oulined above, the semantic analysis of the 

predicates will have the form of an underlying predication.

The Latin predicate delibero  pertains to the semantic field of verbs of 

cognition. Within this vast semantic field, this verb will be classified under 



50

the dimension1 “to think carefully”. The Old English predicate ¥encan, on the 

other hand, represents the archilexematic content of the semantic field. This 

simple difference will play an important role in determining the “best2

possible translation”.

More specifically, the pair delibero/Íencan is a case of non-lexical equiv-

alence between the source and the target language. In this circumstance, as 

van der Korst (1990:291) remarks:

When a word (predicate) of the source language corresponds to 

more than one word (predicate) of the target language, a choice will 

have to be made. In fact, this is the crux in (machine) transla tion.

Indeed, the Latin verb delibero  does not show a one-to-one correspon-

dence with the Old English Íencan. In these cases, the predicates might differ

in various respects:

- qualitative valency: Semantic function assignment3;

- quantitative valency: Syntactic function assignment4 and

- selection restrictions5

1 The organization of the vocabulary in semantic fields follows the tenets of Lexematics 
(Coseriu 1981). For the introduction to the term dimension, see Geckeller (1971).

2 By best we refer to the most approximate translation once we have revised the 
linguistic in formation provided by the underlying predications.

3 The assignment of semantic functions has been a source of dispute among the various 
linguistic schools. However, in order to avoid terminological tedium, we will remain 
faithful to the terminology presented in Dik (1989), who draws on Fillmore (1968, 
1971 and 1977). Studies such as Chomsky (1981, 1986a and 1986b), Halliday (1985) 
etc. present a different picture. 

4 We follow Dik (1979 and 1989) with respect to syntactic function assignment: the 
syntactic functions Subject and Object are assigned at the level of the predication 
formation component rather than at both the level of the predication formation and 
expression components, as Conolly (1991) has put forward.

5 Selection restrictions have also been alluded under the name of contextual restrictions.



51

However, the difference does not necessarily stem from the syntagmatic 

axis alone but, conversely, this might be codified in the paradigmatic organi-

zation. This leads to formulate what we call the Scalarity Principle:

(3) SCALARITY PRINCIPLE: Those predicates higher on the scale 

(paradigmatic) tend to take an ample range of complementation patterns as 

well as a wide semantic scope whereas those lower on the scale (syntagmatic) 

tend to have a reduced ammount of patterns; and, accordingly, its semantic 

scope is narrower.

Thus, delibero  and Íencan:

(4)
a. deliberov (x1: human (x1))Exp/S (x2: PPde <animate-concrete’>

(x2))Go

De aliqua re deliberare

“To discuss something”

b. deliberov (x1: human (x1))Exp/S (x2: NP <‘animate-concrete’>

(x2))Go/O

Re deliberata, post diem tertium ad Caesarem reversuros1

“Having discussed the matter, they would go back to Ceasar’s 

positions after three days”

(5)
a. Íencanv (x1: human (x1))Exp/S

Agunnon Íencan Ía bocerns

“Coeperunt cogitare scribæ”1

1 In Lewis and Short (1966:236).



52

b.Íencanv (x1: human (x1))Exp/S (x2: [p3 X1] (x2))Go/O

Secge he hwæt ic Íence

“Let him say what my thoughts are”2

c. Íencanv (x1: human (x1))Exp/S (x2: PPon <‘animate-concrete’>

(x2))Go

God ys on Dryhten to Íenceanne

“Bonum est confidere in Domino”3

d. Íencanv (x1: human (x1))Exp/S (x2: [p3 X1] (x2))Go/O

Hie Íohton hu hie hine acwellan meahton4

“They were thinking how to kill him”

e. Íencanv (x1: human (x1))Exp/S (x2: [p3 X1] (x2))Go/O

Nænig heora Íohte, Íæt he Íanon scolde gesecean folc5

“Nobody thought that he would search for his people afterwards”

Since we are concerned with the predication-formation component we do 

not draw a distinction between cases like (6) and (7):

(6)

Him Ías Íing Íencendum

“Hæc eo cogitante”6

(where the subject of thought is in the accusative)

1 An Etymological Dictionary of the English Language, in Bosworth and Toller 
(1964:1046)

2 The Blickling Homilies 181, in Bo sworth and Toller (1964:1046)
3 Libri Psalmorum  117:8, in Bosworth and Toller (1964:1046)
4 Libri Psalmorum  241:8, in Bosworth and Toller (1964:1046)
5 Libri Psalmorum  123:4, in Bosworth and Toller (1964:1046)
6 The Gospel according to St. Mathew I:20, in Bosworth and Toller (1964:1046).



53

(7)

We moton Íæs Íencan Íe egeslic on Íissum bocum is gewrit ten1

“We must think about the threatening things that are written in 

this book”

(in these cases the subject of thought is in the genitive)

It is the expression rules, concieved as a bridge between the underlying 

predications and the actual linguistic expressions, that account for case mark-

ing (such as accusative vs. genitive in (6) and (7)).

This analysis reveals that Íencan does not correspond to delibero  but to 

cogito, which shows the same semantic behaviour as Íencan. The first obvi-

ous difference lies in the fact that both predicates differ in their quantitative 

valency. This proves our Scalarity Principle (cf. above).

Our analysis in terms of the Scalarity Principle is also reinforced if we take 

into account the other two lexemes corresponding to the archilexeme Íencan

in Old English: hogian and hycgan:

(8) hogianv (x1: human (x1))Exp/S (x2: PPymbe <‘entity’> (x2))Go

He lythwon hogode ymbe his sawle Íearfe

“He thought little about the needs of his soul” 2

(9) hycganv (x1: human (x1))Exp/S (x2: PPymbe <‘entity’> (x2))Go

HycgeÍ ymbe se Íe wille

“He shall think about he who will” 3

1 Laws of King Athelstan, in Bosworth and Toller (1964:1046).
2 An Anglo -Saxon Reader in Prose and Verse 101, 201 in Bosworth and Toller 

(1964:549).
3 The Anglo -Saxon Version of the Metres of Boethius 19,2 in Bosworth and Toller

(1964:578).



54

As is shown in (8) and (9) Íencan, hogian and hycgan differ with respect 

to their quantitative valency, that is the choice is defined by syntax. These 

verbs, however, also have clausal complementation, although with a differe nt

meaning:

(10) hogianv (x1: human (x1))Exp/S (x2: [p3 X1] (x2))Go/O
We sceoldon hogian hu we hi begyton
“We must consider how we may obtain it”

1

(11) hycganv (x1: human (x1))Exp/S (x2: [p3 X1] (x2))Go/O

Uton we hycgan hwær we ham agen and Íonne geÍencan hu we 

Íider cumen

“Let us consider where we may have a home and then devise how 

we may come thither”
2

Examples (10) and (11) do not contradict our explanation in terms of the 

Scalarity Principle: although one may draw the conclusion that the different 

quantitative valencies that Íencan and hogian and hycgan have cannot ac-

count for the restrictions imposed on the final choice, the fact that these two 

verbs have a different meaning when they undergo clausal complementation 

reinforces our Scalarity Principle explanation: there is a difference of contex-

tual value and the two predicates lose the archilexematic meaning of Íencan

and acquire a more specific meaning that belongs to another dimension3 (“to 

think carefully”) within the same semantic field: to consider.

3.2. amitto/gewitten

1 The Homilies of Ælfric II 316, 25 in Bosworth and Toller (1964:549).
2 Apollonius of Tyre 83a, in Bosworth and Toller (1964:578).
3 Following Coseriu (1981).



55

The second pair of predicates that we examine is the one formed by 

amitto and gewitten. This equivalence involves two fixed expressions, vitam

amittere and gewittan of life . Since there exist a L2 equivalence the steps that 

van der Korst (1989:297) calls paraphrasing and deparaphrasing can be 

avoided:

(12)
amittov (x1: human (x1))Exp/S (x2: NP <‘entity’> (x2))Go/O
Decius amissit vitam1

“Decius passed away”

1 In Lewis and Short (1966:109). We also draw on Gaffiot (1934).



56

(13)
gewittanv (x1: human (x1))Exp/S (x2: PPof <‘entity’> (x2))Loc

He of life gewat

“He departed this life”1

Nevertheless, the fact that there is an equivalence in the target language, 

does not prevent this equivalence from showing a different complementation 

pattern: an NP (nominal complementation) in Latin and a PP (prepositional 

complementation) in Old English.

Therefore, the difference between the two underlying predications can be 

said to be one of qualitative valency.

The scalarity principle proves again useful in this translation procedure 

because these two predicates , which are lower on the scale (syntagmatic), 

have only one pattern of complementation and, accordingly, its semantic 

scope is narrower: to die.

4. CONCLUSION

In this paper we have presented the outlines of a predicate comparison 

procedure of translation in terms of the Scalarity Principle, which sets the 

complementation patterns of a given predicate in the syntagmatic as well as 

in the paradigmatic axis and relates the complementation patterns to the se-

mantic scope of the predicate under scrutiny. This  principle has proved valid 

for a translation procedure by application to two Latin predicates and their 

Old English equivalents.

Ricardo Mairal Usón (Universidad Autónoma de Madrid)
Javier Martín Arista (Universidad de Zaragoza)

1 Beowulf 4934, in Bosworth and Toller (1964:470).



57

REFERENCES

BACH, E. and T. Harms eds. 1968. Universals in linguistic theory. New York: 
Holt, Rinehart and Winston.

BOSWORTH, J. and T.N. Toller (1898)1964. An Anglo-Saxon Dictionary.
Oxford: Oxford University Press.

CHOMSKY, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.

CHOMSKY, N. 1986a. Knowledge of Language. Its Nature, Origin and Use.
New York: Pantheon.

CHOMSKY, N. 1986b. Barriers. Cambridge (Massachusetts): MIT Press.

COLE, P. and M. Sadock eds. 1977. “Grammatical relations” Syntax and 
Semantics 8. New York: Academic Press.

CONOLLY, J. H. 1991. Constituent Order in Functional Grammar: Syn-
chronic and Diachronic Perspectives. Berlin and New York: Foris 
Publications.

CONOLLY, J.H. and S.C. Dik eds. 1989. Functional Grammar and the 
Computer. Dordrecht: Foris.

COSERIU, E. 1981. Principios de semántica estructural. Madrid: Gredos.

DIK, S. C. 1979. Functional Grammar. Dordrecht: Foris.

DIK, S. C. 1986. “Two papers on the computational application of Functional
Grammar” (Working papers in Functional Grammar 18). Ams terdam:
University of Amsterdam.

DIK, S. C. 1989. The Theory of Functional Grammar Part I: The Structure of 
the Clause. Dordrecht: Foris



58

DIK, S.C. and K. Hengeveld: 1990. “The hierarchical structure of the clause 
and the typology of perception verb complements” (Working papers 
in Functional Grammar 37). Amsterdam: University of Amsterdam.

FILLMORE, C. 1968. “The case for case” in Bach and Harm eds. 1968:1-88.

FILLMORE, C. 1971. “Types of lexical information” in Steinberg and Ja-
cobovits eds. 1971:370-392.

FILLMORE, C. 1977. “The case for case reopened” in Cole and Saddock eds. 
1977:59-81.

GAFFIOT F. 1934. Dictionarie Illustré Latin-Francais. Paris: Librairie Ha -
chette.

GECKELLER, H. 1971. Strukturelle Semantik und Wortfeldtheorie translation
by Marcos Martín Hernández (1976) Semántica estructural y teoría 
del campo léxico. Madrid: Gredos.

GOOLDEN, P. ed. 1958. The Old English Apollonius of Tyre. London: Oxford
University Press.

HALLIDAY, M.A.K. 1985. An Introduction to Functional Grammar. London
and New York: Arnold.

HENGEVELD, K. 1987. “Clause structure and modality in Functional
Grammar” in Van der Auwera and Goosens eds. 1987:53-66.

HENGEVELD, K. 1988. “Illocution, mood and modality in a functional
grammar of Spanish”. Journal of Semantics 6.3/4 (1988):227-269.

HENGEVELD, K. 1989. “Layers and operators”. Journal of Linguistics
25.1(1989): 127-157.

LEWIS, C.T. & C. Short (1879) 1966. A Latin Dictionary. Oxford: Clarendon
Press.

LYONS, J. 1977. Semantics (2 vols). Cambridge: Cambridge University Press.



59

STEINBERG, D.D. and L.A. Jakobovits eds. 1971. Semantics, an interdisci-
plinary reader in philosophy, linguistics and psychology. London: 
Cambridge University Press.

VAN DER AUWERA, J. and L. Goosens eds. 1987. Ins and outs of the pred-
ication. Dordrecht: Foris.

VAN DER KOST, B. 1990. “Functional Grammar and Machine Translation” in 
Conolly and Dik eds. 1989:289-316.

* * *