ATLANTIS 25.1 (June 2003): 29-38
ISSN 0210-6124
Lexical and Non-Lexical Linguistic Variation
in the Vocabulary of Old English
Javier E. Díaz Vera
Universidad de Castilla-La Mancha
JavierEnrique.Diaz@uclm.es
This paper deals with different theoretical aspects of lexical variation and change. More
exactly, I will focus here on the analysis of some of the different ways lexical and non-lexical
linguistic variation can interact over long periods of time. My analysis is based on the
Functional-Lexematic Model, which conceives the lexicon of a language as a grammar. Its
central unit of description is the word, which appears with all its syntactic, morphological,
semantic and pragmatic properties. Macrostructurally, predicates are interconnected by
cohesive, associative and encyclopaedic functions, forming what has been called the
semantic architecture of the lexicon. I will deal here with three basic concepts of lexical
analysis and their applications to the historical vocabulary of English: (1) word-frequency,
(2) syntactic variation, and (3) lexical productivity. I will argue that the relative position of
an Anglo-Saxon lexeme within the semantic architecture of OE can be calculated in terms
ofat leastthe following three diffferent types of variation: onomasiological, syntactic and
morphological-derivational. Since defining a verb means locating it in semantic space, the
resulting hierarchical ordering into lexical domains and subdomains implies a reliable
reconstruction of the mental lexicon of the Anglo-Saxons. In doing so, I will try to show how
these easily observable variational phenomena can be used in order to compensate for the
shortcomings of historical lexicology and semantics.
1. Introduction
In this paper, I am going to analyze some of the different ways lexical and non-lexical types
of linguistic variation interact over long periods of time, with special attention to the early
history of English.1 My approach is mainly lexicographic, in that this research is based on
day-to-day observation and analysis of the information contained in different types of
dictionaries of the English language, both historical and non-historical, semasiological and
onomasiological, bilingual and monolingual, general and specialized.
In fact, the primary source of inspiration for this paper is to be found in my earlier
work on the compilation of a formalized grammatical lexicon of Old English (henceforth
OE) and Old Norse verbs, organized onomasiologically in semantic hierarchies (Díaz Vera
2002). Within this ongoing project, we are trying to apply some of the methodological
principles of the so-called Functional-Lexematic Model (hence FLM) (Faber and Mairal
1. This paper is part of the research project GermaLex: Diccionario onomasiológico contrastivo
del léxico verbal de las lenguas germánicas antiguas (MCyT BFF2001-0921).
30
Javier E. Díaz Vera
Usón 1999) to the analysis of the historical lexicon of English. In order to do so, we have
adapted some of the basic principles of lexical description formulated by the FLM to the
analysis of the lexicon of past states of language. Through the careful observation of the
overall functioning of the verbal lexicon of Present-Day English (henceforth PDE), we
have been able to formulate a set of theoretical statements whose application to our
lexicographic approach to the vocabulary found in Anglo-Saxon texts has proved
extremely useful. To put it in Labovian terms, my main interest is in how to use the
present to explain the past, so that I will focus on the application of the theoretical findings
of the synchronic version of the FLM to the lexical analysis of past states of language.
2. Lexical and non-lexical variation in the FLM, synchronic and diachronic
To start with, I will briefly introduce some of the basic principles of the synchronic version
of the FLM, with special attention to its concept of linguistic variation. In the FLM, the
lexicon of a language is conceived of as a grammar. Its central unit of description is the
word, which appears with all its syntactic, morphological, semantic and pragmatic
properties (Faber and Mairal Usón 1999: 57). Macrostructurally, predicates are
interconnected by cohesive, associative and encyclopaedic functions, forming what has
been called the semantic architecture of the lexicon (Faber 1994; Díaz Vera 2002).
The term lexical domain is used to refer to the highest level of onomasiological
organization within the FLM. A lexical domain is constituted by the set of lexemes which
together lexicalize all or part of a conceptual domain. Lexical domains are further organized in a network of subdomains, i.e. meaningful sets of lexemes within a lexical domain,
which form the basis of its internal structure and focus on a particular area of meaning.
For example, the set of English verbal predicates included under the lexical domain of
possession to have something (e.g. have) can be grouped together in smaller subdomains
or subareas of meaning, such as to come to have something (e.g. get, obtain), to
continue to have something (e.g. keep, save), to stop having something (e.g. lose) or to
cause somebody or something to have something (e.g. give). In turn, these subdomains
may be further divided into smaller semantic areas.
Each domain or subdomain has a superordinate term, the genus, by means of which
its troponyms are directly or indirectly defined. Following the previous example, we can
affirm that the ultimate genus of the English domain of possession is the verb to have,
which will appear in a more or less direct way in the individual definitions of each lexical
item within this domain. Similarly, to get is the genus of the subdomain to come to have
something, so that every lexical member of this lexical group will we defined by means
of it (e.g. take, gain, receive, collect, recover). Furthermore, these predicates instantiate new
lexical subdomains, whose members are defined in terms of a common genus. Meaning
definitions include a second type of semantic information, referred to as differentiae,
which distinguishes a lexeme from others in the same lexical domain.
2.1. Onomasiological variation: the role of word-frequency in historical lexicography
I will now concentrate on the set of relationships that can be established between a genus
and its troponyms in terms of lexico-semantic and morphosyntactic variation. To start
Lexical and Non-Lexical Linguistic Variation in the Vocabulary of Old English
31
with, as has been noted above, the semantic area covered by a genus must include all its
troponyms, so that semantic predicates with archilexematic status are the most general
ones within their corresponding categories. For examples, PDE to get includes in its
definition such predicates as to gain, to buy or to obtain, whereas to obtain includes the
more specific terms to procure and to acquire. Broadly speaking, we can argue that, whereas
the use of the most specific terms within each category depends frequently on stylistic and
sociolinguistic factors, archilexematic predicates are found much more frequently in
general language. In order to illustrate this tendency, I will analyze the distribution of two
couples of near-synonyms in the Brown Corpus (which includes samples of 15 categories
of American English texts from the early 1960s). The lexical couples under scrutiny are
composed of PDE to obtain to get something as a result of work/planning plus one of
its troponyms, either to acquire or to procure. As can be seen here, PDE to obtain is used
much more frequently than to acquire in all the genres represented in this corpus.2 The
relative frequency of PDE to acquire is relevant (that is, over 40%) only in two of the
textual categories analyzed here, skills and hobbies and belles lettres. In order to interpret
this distribution, I will argue here that all the texts included under these two categories
have in common their reader-oriented character. Broadly speaking, these texts consist of
instructional treatises, so that the writers message must be efficiently decoded by the
reader. This implies that different lexemes will be used in order to express apparently
insignificant semantic nuances, avoiding polysemy and ambiguity.
Press
Learned
Government
Religion
Popular
Lore
Skills and
hobbies
Belles
lettres
to
obtain
0.82
[23]
0.81
[70]
0.79
[31]
0.67
[2]
0.65
[15]
0.59
[13]
0.50
[12]
to
acquire
0.18
[5]
0.19
[16]
0.21
[8]
0.33
[1]
0.35
[8]
0.41
[9]
0.50
[12]
The analysis of the couple to obtain/to procure is still more revealing, in that it shows
clearly that the use of the more specific predicate is restricted to the most reader-oriented
category within this set.
Press
Learned
Government
Religion
Popular
Lore
Skills and
hobbies
Belles
lettres
to
obtain
1.00
[23]
1.00
[70]
0.97
[31]
1.00
[2]
1.00
[15]
1.00
[13]
0.70
[12]
to
procure
0.00
0.00
0.03
[1]
0.00
0.00
0.00
0.30
[5]
These two calculations indicate that, given a corpus of general language use, archilexematic predicates are used more frequently than their troponyms. More importantly,
2. Only in the set of texts labelled as Belles lettres these two verbs are used with the same
frequency.
32
Javier E. Díaz Vera
the selection of the more specific predicates by the speaker or writer seems to correlate
with textual factors, allowing a variationist approach to this type of onomasiological
variation.
The word-frequency criterion has influenced OE lexicography in different ways. On
the one hand, a preliminary application of this basic principle can be found in Roberts and
Kays Thesaurus of Old English (TOE, 1995). Using the information given by dictionaries
of OE, the compilers of the TOE have included simple references to word-frequency (very
infrequent words are flagged with o) and textual category (flag p for words that appear
only in poetry, and flag g for words used generally as glosses of Latin texts). On the other
hand, word-frequency has become one of the overriding principles in the editing of the
Dictionary of Old English (Amos and Healy 1986).
In our lexicographic approach to OE, word-frequency is used as a preliminary
indicator of archilexematic status, so that it is applied exclusively to couples or sets of
lexemes belonging to the same lexical subdomain, and traditionally treated by modern
lexicographers of OE as synonyms or near-synonyms. What we argue is that, given a set
of OE lexemes that form a lexical subdomain, the most general predicate will tend to
appear more frequently than its troponyms and its usage will be more or less regular
through different textual types. By way of illustration, we can have a look at the following
set of OE verbs of light, which the compilers of the TOE (141) have arranged alphabetically as a compact set of synonyms:
To shine: ascinan, beorhtian, (ge)bierhtan, blic(i)ang, gladian, (ge)lihtan, lymano,
onlihtan, (ge)scinan
Using the Dictionary of Old English Corpus, I have calculated the relative frequencies of
these nine lexical items. According to my calculation, OE (ge)scinan is by far the most
frequently used verb of light in OE texts (143 occurrences, corresponding to 38.23%),
whereas, as one might expect, the flagged items OE blic(i)ang and OE lymano occupy the
lowest positions (with one single instance each, amounting to 0.26%). According to our
observations of the overall functioning of PDE predicates, we could argue that OE scinan
is the most general verb within this group, acting as the genus in the definitions of the
remaining predicates of light.
However, the application of this onomasiological principle to historical lexicography
is not free from complications, insofar as our corpus of OE is small and probably not
representative enough of the lexicon of the Anglo-Saxons. In spite of this, rather than
denying the validity of word-frequency as a lexical research tool for the calculation of
degrees of semantic coverage, I will try to show here that, by analyzing and combining
different types of linguistic variation in our limited corpus of Anglo-Saxon texts, we can
get a highly reliable picture of the hierarchical onomasiological structure of OE.
2.2. Syntactic variation and lexical membership
As has been seen before, all the predicates included in the same lexical subdomain can be
defined in terms of the same genus. This tendency is formulated in the so-called Principle
of Lexical Domain Membership (Faber and Mairal Usón 1999: 87):
Lexical and Non-Lexical Linguistic Variation in the Vocabulary of Old English
33
Principle of Lexical Domain Membership: Lexical domain membership is determined by the genus, which constitutes the nucleus of the meaning of a lexeme
In order to apply this principle to the lexical analysis of OE, it should be remembered here
that our interpretations of the meaning of each individual lexical item must rely almost
exclusively on the definitions developed by previous lexicographers of that language.
Contrary to PDE dictionaries, OE dictionaries (e.g. Bosworth and Toller 1898, Hall 1960)
do not normally offer full definitions of each lexical item, from which one could extract
different types of semantic and grammatical information. Rather, historical dictionaries
of English can be described as bilingual dictionaries, insofar as the lexicographic definitions they include are simple approximate translations of each OE lexical item into
Modern English.
In the TOE, Roberts and Kay have used the translations offered by Old English
alphabetical dictionaries in order to propose an onomasiological arrangement of the bulk
of Anglo-Saxon lexemes. By turning Halls A Concise Anglo-Saxon Dictionary (1960) and
its supplements inside out (Roberts and Kay 1995: xvi), the compilers of the TOE have
assigned OE lexemes to semantic groups and subgroups, following a mechanism of
semantic clustering that vaguely resembles the principle of lexical domain
membership described above. However, whereas in the TOE words are ascribed to
semantic areas on the basis of their sharing any meaning component, the FLM postulates
that lexemes can be grouped on the basis of a common core meaning.
Given the limits that result from a paradigmatic analysis alone in historical lexicology,
linguistic variation can be used as a source of information on the configuration of lexical
domains and subdomains. More exactly, I will analyze now some of the different ways
syntax can be used in order to throw light on some problematic issues related to the
classification of OE verbal predicates into lexical subdomains. Within the FLM, it has been
observed that the members of each lexical subdomain are interrelated not only by means
of their common genus, but also by a similar syntactic behaviour, i.e. they share one or
more syntactic complementation patterns. The syntax of each individual predicate is
conceived of as being motivated by the lexical subdomain to which the predicate in
question belongs.
For example, the onomasiological arrangement proposed in the TOE for OE verbs of
sensation, perception, feeling (02.05, 6479) includes the subdivisions shown in Table 1
below. Although the existence of semantic connections between these ten semantic
categories is quite evident, our syntactic analysis of a representative number of predicates
from each category indicates that these semantic links are not strong enough to be
conceived a compact lexical domain. To start with, OE verbs of perception share at least
one of the two basic syntactic complementation patterns inherited from ProtoGermanic,
namely NPACC and NPGEN (with a Human participant in the role of subject/experiencer).
This implies that all the intransitive predicates in 02.05.01 to 02.05.05 can be automatically
excluded from this category, so that our lexical domain perception will be composed of
five basic subdomains, corresponding to the five human senses: touching, tasting, smelling,
seeing and hearing. Furthermore, TOE 02.05.10 faculty of hearing includes the
subcategories in Table 2. Rather than focusing on the faculty of hearing, this is a classification of different types of sounds, from uproar to whistle. Once again, the semantic link
34
Javier E. Díaz Vera
between noise and the faculty of hearing is more than obvious. However, it should be
noticed that most of the verbs included under these headings do not admit any of the two
syntactic patterns described above for OE verbs of perception, so that, in our opinion,
they should be included in another lexical domain, i.e. sound.
02.05. Sensation, perception, feeling
02.05.01. Without feeling, insensible
02.05.02. Consciousness
02.05.03. Exhaustion, faintness, weariness
02.05.04. Sleepiness, drowsiness, sleep
02.05.05. Desire, appetite
02.05.06. Sense of touch
02.05.07. Faculty of taste
02.05.08. Faculty of smell
02.05.09. Faculty of sight
02.05.10. Faculty of hearing
Table 1
02.05.10
Faculty of hearing
02.05.10.01. Thing heard
02.05.10.02. Noise, din
02.05.10.03. Noise, tumult, uproar
02.05.10.04. To resound
02.05.10.05. Roaring, raging
02.05.10.06. Harsh sound
02.05.10.07. A crashing, noise
02.05.10.08. A crackling, rustling
02.05.10.09. A hissing, whistling
02.05.10.10. Clattering, noise
Table 2
2.3. Syntactic variation and semantic coverage
In the FLM, defining a predicate means locating it in semantic space. Once lexical items
have been grouped into lexical domains and subdomains, we must be able to reconstruct
their relative position within the corresponding category, elaborating lexical hierarchies
where semantically related lexemes are ordered according to their semantic space and
relative location within a subdomain.
The analysis of the verbal lexicon in PDE leads to an interesting conclusion, that the
hierarchical semantic ordering of lexemes correlates with a systematic distribution of their
syntactic configuration. This basic principle of the synchronic version of the FLM (Faber
and Mairal Usón 1994: 21011) has been formulated in terms of the so-called Lexical
Iconicity Principle:
Lexical and Non-Lexical Linguistic Variation in the Vocabulary of Old English
35
Lexical Iconicity Principle: The greater the semantic coverage of a lexeme is, the
greater its syntactic variation.
Syntactic variation is to be understood here as the number of complementation patterns
admitted by a verbal predicate. The superordinate term within each subdomain tends to
take a greater number of complementation patterns than its more specific troponyms.
Cortés Rodríguez and Mairal Usón (2002) have adapted this principle to the analysis
of the lexicon of past states of the language, proposing the following reinterpretation of
the lexical iconicity principle (provisionally called Beta-Reading 2021):
Lexical Iconicity Principle (Beta-Reading): The greater the syntactic coverage
of a lexical unit, the higher its position in the semantic hierarchy within a given
subdomain
This new interpretation introduces a reorganization of both the semantic and the
syntactic component in the sense that it is now syntax that determines the semantic space
and location in a subdomain (21). According to this new principle, we can expect that:
1. Syntactic variation compensates for the shortcomings of a semantic analysis.
2. Syntactic patterning allows us to articulate not only configurational structure
within domains, but also to assign verbal predicates to lexical classes.
Let us have a look, for example, at the following list of OE verbs of touching (Díaz
Vera 2002). As we can see from their dictionary translations into Modern English, these
predicates are treated as a compact group of synonyms with the general meaning to
touch (see also TOE 68, for the resulting onomasiological arrangement):
Bosworth and Toller (1898)
Hall (1960)
grapian
To grope, touch, feel with the hands
To touch, grope
hrepian
To touch
To touch
hrinan
To touch
To touch
tacan on
To touch (Toller 1921)
To touch
(ge)tillan
To touch
To touch
Table 3: Dictionary translations of OE verbs of touching
In order to rearrange these five verbs according to their degree of salience, trying to
establish which word was preferred by OE speakers to refer to this particular action, I have
calculated the degree of syntactic variation of each predicate with the results shown in
Table 4 below.
My analysis clearly indicates two different things. Firstly, it has been seen that OE verbs
of touching share at least one of the two basic complementation patterns, NPACC or PP,
36
Javier E. Díaz Vera
which act as indicators of lexical membership for this set of words. Secondly, given its
higher degree of lexical variation, we have been able to identify OE hrinan as the most
general predicate within this lexical group, so that it can be considered the subdomain
genus.
Verb
hrinan
No.
of
cps
4
Inventory
of cps
Examples
NPACC
[PPs: 085900 (104.13)] Ne sceolon ge mine þa halgan hrinan
lit. You should not touch my holy god
NPGEN
[GenA,B: 022100 (614)] Nu þu his hrinan meaht.
lit. Now you may touch it.
NPDAT
[MtGl (Ru): 067300 (20.34)] Se hælend & hran egum heora
lit. The Saviour touched their eyes
PPæt
[Beo: 062500 (2267)] Oþ ðæt deaþes folm hran æt heortan
lit. Until the hand of death touched at the heart
o/
[Beo: 057700 (2081)] He mægnes rof min costode, grapode
gearofolm
lit. he made proof of me, groped out ready-handed
NPACC
[ÆCHom II, 10: 001400 (82.39)] Se cuma his cneow
grapode. mid his halwendum handum
lit. The stranger touched his knee with his healing
hands
grapian
2
hrepian
1
NPACC
[ÆCHom I, 8: 002100 (242.30)] Moyses æ forbead to
hreppene ænigne hreoflan
lit. The law of Moses forbade to touch any leper
(ge)tillan
1
NPACC
[BenR: 019800 (7.23.1)] Gif we þone hrof ðære healican
eaðmodnesse getillan willaþ
lit. If we want to touch the roof of holy humility
tacan
1
PPon
[LS 29 (Nicholas): 011200 (273)] Sona swa þæt ele toc on
þæt wæter, þa aras þær upp swiðe mycel fyr
lit. As soon as the oil touched on the water, there
arose a great fire
Table 4: OE verbs of touching: degrees of syntactic variation.
2.4. Lexical productivity
Our reconstruction of different lexical domains and subdomains in OE has led to other
interesting findings about lexical organization and onomasiological salience. To start with,
I will make reference to what I have called the Lexical Derivational Principle, which
implies that, broadly speaking, the higher positions within a given semantic hierarchy are
normally occupied by underived verbs that frequently serve as a basis for the creation of
new derived words. This tendency has been captured in the following principle:
Lexical and Non-Lexical Linguistic Variation in the Vocabulary of Old English
37
Lexical Derivational Principle: The greater the semantic coverage of a lexeme is,
the greater its number of derivational formations (Díaz Vera 2002: 5556).
The rationale behind this definition is as follows. A predicate is onomasiologically stronger
than its hyponyms if it is a likely choice for the derivation of new lexemes within its lexical
subdomain. I will now apply this principle to our previous list of OE verbs of touching.
In order to calculate the relative degree of lexical productivity of a lexeme, I have made the
following list, which includes all the lexemes derived from these five predicates:
Lexical productivity (lp)
Noun
Adjective
OE verb
Verb
grapian
agrapian
grapung
grapigendlic
5
hrepian
ahrepian
hrepung
ungehrepod
4
ahrinan
andhrinan
æthrinan
gehrinan
onhrinan
oþhrinan
hrine
hrining
æthrine
handhrine
hrinenes
gehrinenes
onhrine
ungehrinen
14
0
2
9
3
25
hrinan
tacan
tillan
Total
atillan
getillan
13
Total
Table 5: OE verbs of touching: degrees of lexical productivity
The degree of lexical productivity of a lexeme is the number of lexemes derived from it
divided by the cumulative number of derived lexemes calculated for all the members of its
same lexical subdomain. Thus, given our corpus of OE texts, the lexical productivity
degree of an item like OE hrinan can be calculated by counting how many times it has
been used to create new words (i.e. 14) and then comparing our result to the total number
of lexemes derived from the whole set of predicates that form the Old English subdomain
of touching (i.e. 25, so that 14/25 = 56.00%). In fact, one could claim that as long as we
move down the semantic scale, from the most general to the most specific term, the
number of semantic specifications that can be expressed through lexical derivation from
a single lexical root decreases (Díaz Vera 1999: 80).3
3. The correlation between onomasiological variation, frequency of usage and number of derived
word-forms has been ascertained in other languages. For example, Inoue (2001: 44-72 and personal
communication) demonstrates that, according to the evidence provided by the Linguistic Atlas of
Japan, words with a higher frequency of usage are preferred by speakers of Japanese in order to
derive new words.
38
Javier E. Díaz Vera
3. Conclusions
According to the principles defined above, the relative position of an Anglo-Saxon lexeme
within the semantic architecture of OE can be calculated in terms of, at least, the following
three different types of variation: onomasiological, syntactic and morphologicalderivational. Since defining a verb means locating it in semantic space, the resulting
hierarchical ordering into lexical domains and subdomains implies a reliable reconstruction of the mental lexicon of the Anglo-Saxons.
Much more importantly, I have tried to show how these easily observable variational
phenomena can be used in order to compensate for the shortcomings of historical
lexicology and semantics.
Works Cited
Amos, A. C., and A. P. Healey 1986: Dictionary of Old English. Toronto: U of Toronto P. (Continuing
microfiche publication.)
Bosworth, Joseph, and T. Northcote Toller 1972 (1898): An Anglo-Saxon Dictionary. London: Oxford
UP.
Cortés Rodríguez, Francisco J., and Ricardo Mairal Usón 2002: A Preliminary Design for a Syntactic
Dictionary of Old English on Semantic Principles. A Changing World of Words: Studies in English
Historical Lexicography, Lexicology, and Semantics. Ed. Javier E. Díaz Vera. New York and
Amsterdam: Rodopi. 346.
Díaz Vera, Javier E. 1999: Remembering in Old English: The Diachronic Reconstruction of a Verbal
Dimension. Atlantis 23: 525.
2002: The Semantic Architecture of the Old English Verbal Lexicon: A HistoricalLexicographical Proposal. A Changing World of Words: Studies in English Historical Lexicography,
Lexicology and Semantics. Ed. Javier E. Díaz Vera. New York and Amsterdam: Rodopi. 4777.
Faber, Pamela. 1994: The Semantic Architecture of the Lexicon. Symposium on Lexicography IV.
Proceedings on the 3rd. International Symposium on Lexicography. Ed. K. Hyldgaard-Jensen and
V. Hjørnager Pedersen. Tubingen: Max Niemeyer. 3750.
and Ricardo Mairal Usón 1994: Methodological Underpinnings for the Construction of a
Functional Lexicological Model. Miscelánea 15: 193217.
and Ricardo Mairal Usón 1999: Constructing a Lexicon of English Verbs. Berlin and New York:
Mouton.
Hall, J. R. Clark 1960: A Concise Anglo-Saxon Dictionary. Toronto: U of Toronto P.
Inoue, Fumio 2001: Keiryoteki Hogen Kukau [Quantitative Dialect Classification]. Tokyo: Meiji
Shoin.
Roberts, Jane, and Christian Kay 1995: A Thesaurus of Old English. London: Centre for Late Antique
and Medieval Studies.
Toller, T. Northcote 1921: An Anglo-Saxon Dictionary: Supplement . Oxford: Oxford UP.
