5.1 Introductory Remarks
In the previous chapter we reported the results of a quantitative study of the rise of definite and indefinite articles in North Germanic. While the statistical data give us some picture of the grammaticalization processesâfor instance, demonstrating that the grammaticalization of the definite article began with topicalized discourse referentsâa qualitative analysis of the corpus is needed to elaborate on and clarify the picture. In the present chapter we will therefore return to the models of grammaticalizations of both articles which were presented in Chapter 1. We begin with a discussion of the model of grammaticalization of the definite article, and its limitations, with particular attention to the developments from one stage to the next, in particular the unclear path from direct to indirect anaphora as well as from indirect anaphora to larger situation use. We argue that these steps can be explained in terms of referent accessibility, and will give examples to support our claims. We then turn to the model of grammaticalization of the indefinite article and its applications. A proposal encompassing both models will be discussed in 5.5.
5.2 On Models of Grammaticalization
We have noted earlier that there remain unsatisfactory gaps in the proposed model of grammaticalization of the definite article (see section 1.1). For instance, it is still not entirely clear which uses of the definite article constitute the so-called bridging contexts (in the terminology of Heine 2002) between stage I and stage II of the grammaticalization (direct and indirect anaphoric use). In Chapter 4, we have noted that the proposed order of events seems to be confirmed in terms of the frequency of the incipient article; the form does appear earlier and with higher frequency as a marker of direct anaphoric reference, later as a marker of indirect anaphoric reference, and finally in larger situation use (see section 4.8). However, one of the purposes of the present chapter is to find examples which may illustrate the path from direct to indirect anaphor and from indirect anaphora to larger situation use.
Before we continue, we would like to make the following observations: even though the model of grammaticalization presents a chronological development, it is not unusual for different stages of grammaticalization to co-exist at the same time in a languageâs history. In a famous study of the grammaticalization of the future construction âbe going toâ, Fischer and Rosenbach (2000:3) demonstrate how all of the stages of grammaticalization are still present in modern English.
Etymologically, definite articles are derived from deictic elements, often demonstratives. It may happen that the original demonstrative lives on alongside its descendantâthe definite articleâalthough it may be more common that the two differ in form as well as in scope of use. In Icelandic the definite suffix -inn is found together with the demonstrative hinn from which it derives, and both may be used within the same NP; however, the two forms are quite distinct. In Danish and Swedish the demonstrative hinn is now lost.
Indefinite articles seem to stem universally from the numeral âoneâ. Interestingly, very often the numeral and the indefinite article are both retained, sometimes with slight formal differences. In Danish and Swedish the numeral is typically stressed, while the indefinite article is unstressed unless contrast or emphasis is intended.
Irrespective of whether the original demonstrative is retained or lost after definite article grammaticalization has reached more advanced stages, the diachronic connection between the two must at some point come to a conclusion. We may imagine the process of grammaticalization as a gradual extension of the original demonstrative function, which may not be fully visible to users of the language. At some point in the process, however, the form comes to a turning point, from which it may continue as either a demonstrative or a definite article, but not as a form sharing some features of both. In other words, at some point we come to deal with a definite article, a form which may be used in certain contexts in which the use of the demonstrative is not grammatically correct.
The first purpose of a grammaticalization model is thus to establish this turning point in the formâs development, after which the form will be categorized differently than it was before. From this point, the grammaticalization may also proceed even further. Because it is not always easy to decide when a given form has changed categorically, e.g., from a numeral into an indefinite article, in the text we refer to the grammaticalizing forms as incipient or nascent articles. The second purpose is to identify the successive stages of the process, some of which may be located chronologically before the turning point (e.g., the direct anaphoric use of the demonstrative) and some after the turning point (e.g., the use with unique referents). The grammaticalization model should also have a predictive value. It is much more interesting if the model, when applied to previously unstudied languages, indicates that they have gone through the same stages and in the same order.
In the following sections we will discuss the models proposed so far, evaluating them against the data from North Germanic. The empirical data gathered for the sake of the present project allows us greater insight into the process of article formation, and our intention is to incorporate it into the fine-grained models of article grammaticalization.
5.3 Grammaticalization of the Definite Article
5.3.1 From Deixis to Direct Anaphora
The model of grammaticalization adopted here assumes the definite article to derive from a deictic marker, typically a (distal) demonstrative pronoun, as is the case in North Germanic. In this model, the first step in article grammaticalization is the extension of the use of the demonstrative to point not only within a physical situation, but also within a text (anaphora). We have already noted that demonstratives are almost always allowed to be used anaphorically, including when a language has a grammaticalized definite article (in English, for example, this or that can be used to refer anaphorically, even though the definite article the is also available). However, some idiosyncrasies may exist as regards the anaphoric use of demonstratives, for example, a preference to be placed in postposition rather than in preposition, which is the case in Polish (a language without a definite article). Examples (147â148) illustrate the possible locations of the (proximal) demonstrative ten âthisâ in deictic and anaphoric uses.
Deixis:
(147)
a.
Podaj
mi
tÄ
ksiÄ Å¼k-Ä.
pass
me
this.acc.sg
book-acc.sg
âPass me this book.â
b.
?Podaj
mi
ksiÄ Å¼k-Ä
tÄ.
pass
me
book-acc.sg
this.acc.sg
âPass me this book.â (intended meaning)
Anaphora:
(148)
KsiÄ Å¼ka
ta
/
ta
ksiÄ Å¼ka
ilustruje
zmiany
jakie
zaszÅy
w
polskim
spoÅeczeÅstwie.
book.nom
this.nom.sg
/
this.nom.sg
book.nom
depicts
changes
which
took.place
in
Polish
society
âThis book depicts changes which took place in Polish society.â
The Polish examples demonstrate that, while the preposed demonstrative may be used both deictically and anaphorically, the postposition of the demonstrative is allowed only in anaphoric use. Thus, even if the demonstrative cannot be said to be grammaticalizing into a definite article (yet?), its use as an anaphoric marker may be slightly different from its use as a deictic marker.
In the following section, we will investigate the direct anaphora, its markers, and the factors influencing the choice of marker, as well as at anaphoric chains, namely a number of different anaphors with the same antecedent. We will pay particular attention to the so-called Accessibility Marking Scale (Ariel 1988, 1994, 2014), which organizes the potential markers of direct anaphora on a scale, depending on the accessibility of the discourse referent.
5.3.1.1 Accessibility Marking Scale
By direct anaphora we understand a relation between two co-referring linguistic elements: an antecedent and an anaphor. While a typical antecedent is an indefinite NP and a typical anaphor is a pronoun or a definite NP, the form of the anaphor seems to be dependent first and foremost on the accessibility of the discourse referentâthe more accessible the referent is, the less marking is necessary. For direct anaphors, referent accessibility, i.e., how easy it is to find a referent for an anaphor (via its antecedent), is influenced by a number of factors, such as the distance from the antecedent, the number of competitors for the role of antecedent, topicality, and reference frames (domains) relevant for the identification of the antecedent (Ariel 1988:65). The most accessible referents receive zero marking, while the least accessible ones need to be presented by means of full NPs, in Arielâs terminology long definite descriptions. Between these two extremes a number of possible structures are located, arranged in a hierarchy from most to least accessible in Arielâs Accessibility Marking Scale, as presented in (149).
(149) The Accessibility Marking Scale (Ariel 1988)
zero < reflexives < agreement markers < cliticized pronouns < unstressed pronouns < stressed pronouns < stressed pronouns + gesture < proximal demonstrative (+NP) < distal demonstrative (+NP) < proximal demonstrative (+NP) + modifier < distal demonstrative (+NP) + modifier < first name < last name < short definite description < long definite description < full name < full name + modifier
We may note that loss of accessibility is typically compensated for by choosing a fuller form for the anaphor, for example, a definite NP instead of a pronoun. We note further that a simple repetition of the same noun without any modifiers or determiners is not included in the marking scale. The model is thus applicable to an article language, where BNs are not allowed in direct anaphoric contexts. In a language with no grammaticalized definite article the definite descriptions need to include definite determiners other than the article; on the other hand, BNs must also be included in the scale, since it is perfectly natural to use a BN anaphorically in a language with no definite article.
(150)
Samochód
potrÄ ciÅ
ojc-a
przechodzÄ cego
z
córk-Ä
na
pas-ach.
Dziewczynk-Ä
odwiezio-no
do
szpital-a.
Car.nom
hit
father-acc
crossing
with
daughter-instr
on
crossing-instr
Girl-acc
take-impers
to
hospital-gen
âA car hit a father and a daughter who were crossing the street on a pedestrian crossing. The girl was taken to hospital.â
Historically, all article languages are derived from languages without grammaticalized articles; therefore, in a diachronic study BNs should also be considered as part of the Accessibility Marking Scale. For the sake of our study, we disregard pronominal NPs, since we wish to examine different forms of the noun in the anaphoric context. We therefore consider only the variation between BNs, NPs with a demonstrative, and definite NPs in direct anaphoric contexts. We propose a simplified accessibility marking hierarchy in the following form:
(151) BN > NP + (adjectival) modifier > NP + demonstrative (+modifier) > short definite description > long definite description
We assume that among the referents referred to by means of a full NP, the most accessible referents will appear as BNs and the least accessible as long definite descriptions. Our accessibility marking hierarchy is based entirely on Arielâs, with BNs added and only full NPs included (i.e., those with nouns as heads and not pronouns).
We must stress that the extant texts on which our corpus is based already display a regular formal difference between the incipient definite article and the demonstrative from which it has developed (see also section 2.2.2). As early as 1200, the incipient definite article takes the form of a clitic, attached to an inflected form of the noun (and noun alone), while the demonstrative remains a free lexeme. We therefore differentiate between NPs of the type thenne man âthis manâ and mann-in âman-defâ in the oldest texts, and even though we would not claim that the clitic is a definite article as early as 1200, we will treat NPs with -IN as definite descriptions.
5.3.1.2 Direct Anaphors in the Corpus
In the material we find that BNs do appear as direct anaphors, as in (152) where the noun bo âestateâ is consistently presented as a BN.
(152)
Sitær
konæ
i
bo
dör
bonde
kallæ-s
havandæ.
varæ.
hon
skal
i
bo
sittia
tyughu
ukur.
þa
skal
a
seæ
æn
hun
ær
havande.
þa
skal
bo
skiptæ
Sits
woman.nom
in
estate.acc
dies
husband.nom
call-refl
having
be.conj
she
shall
in
estate.acc
sit
twenty
weeks
then
shall
on
see
if
she
is
having
then
shall
estate.acc
divide
âIf a woman is in the estate and her husband dies and she is said to be pregnant she should remain in the estate for twenty weeks, then it will be seen whether she indeed is pregnant. Then, if she is found to be with child, the estate should be divided.â (SV_AVL, Period I)
The noun bo in example (152) is a common noun, used with a specific reference. In Modern Swedish the same noun would never appear as a BN in a direct anaphoric context.
For Period I, when the grammaticalization of the definite article has just began, we would expect to find BNs as anaphors. However, sometimes, when a BN is found in a direct anaphoric context, we find that there are some factors which may favour this NP type over others. Thus in (153), we find the BN kóng-ur âking-nomâ used to refer anaphorically to King Rikardur, who has been presented earlier in the text.
(153)
RÃkharður
kóng-ur
hélt
mikla
skemmtun
á
að
fara
á
skóg [â¦].
Ãess
er
getið
eitt
sinn
að
kóng-ur
var
á
skóg
far-inn.
Rikardur
king-nom
held
great.acc
enjoyment.acc
on
to
travel
to
forest.acc
this.gen
is
said
en
time
that
king-nom
was
in
forest.acc
gone-ptcp
âKing Rikardur greatly enjoyed going to a forest [â¦]. This is said of one time when the king was travelling in a forest.â (IS_Vil, Period III; Piotrowska and Skrzypek 2020)
In this group of examples, we find mostly unique NPs (both more âglobalâ ones such as konung âkingâ, and more âlocalâ ones such as höfding âchieftainâ) and kinship terms. It may be that their accessibility is strengthened by virtue of their relation to the utterance situation and therefore strong enough for no marking to be applied. We must note, however, that this use of BNs is found mainly in Period I and only occasionally later (see section 4.8).
The second group of direct anaphors consists of nouns with adjectival modifiers. These can be of two sorts: the strong form (the so-called indefinite adjective form) and the weak form (the so-called definite adjective form) (see also section 1.2). In modern North Germanic only, the weak form is allowed in definite NPs, while the strong form is combined with indefinite nouns. In the studied corpus such restrictions were not yet established, and strong adjective forms could be found in definite NPs (see especially Delsing 1994). In the material we found few instances of nouns with strong adjectival modifiers but no other determiners. The strong adjective form is found in (154) and (155); however, the latter is a clear example of an anaphoric adjective, meaning âaforementionedâ.
(154)
Tiughu
aar
boþo
þøn
saman
barnløs.
Ok
baþo
guþ
giua
sik
þæt
barn
hanom
mate
þiæna
ok
louaþo
for
oburit
barn
giua
guþ-i
þæt
gærna
tel
þiænist
twenty
years
lived
they
together
childless.
And
prayed
God
give
refl
that
child
him
may
serve
and
promised
for
unborn
child
give
God-dat
it
gladly
to
service
âFor twenty years they lived childless together and prayed to God to give them a child which may serve him and promised to give the unborn child gladly in Godâs service.â (SV_Bur, Period I)
(155)
og
með
umráði
og
samþykki
kórs-bræðr-a
kaus
hann
nú
nefnd-an
ábót-a
til
biskup-s
and
with
consideration
and
approval
cross-brother-gen.pl
appointed
he
now
named-acc
abbot-acc
to
bishop-gen
âAnd by consideration and approval of the brothers he appointed the aforementioned abbot to bishop.â (IS_Arna, Period I)
In the entire corpus we have not found a single instance of direct anaphors consisting of just the noun and the adjectival modifier in the weak form (see also Pfaff 2019 who finds such combinations extremely rare in Old Norse). It seems that, since the anaphoric relation strongly resembles the deictic relation, it is unlikely that an adjectival modifier would suffice to make such a relation clear. The adjectives that we did find were mostly anaphoric (âaforementionedâ, âsaidâ, etc.).
The next NP type is frequently used in direct anaphoric contexts. It consists of nouns with demonstratives, such as denna âthis hereâ (etymologically a compound of demonstrative sá âthisâ and intensifying -si; Skrzypek 2012:64, though most likely no longer transparent in the texts quoted) and den âthisâ. While the former often appears with noun alone (example 156), the latter is usually found with adjectival modifiers. NPs consisting of simple demonstrative and noun alone are more often found in Danish texts (see examples 156 and 157) and in Swedish texts translated from German; this structure is thus typically regarded as a result of German influence (HaskÃ¥ 1972).
(156)
Hoss
thæt
closter
var
haffu-er
oc
eth
marke
var
thære
hoos
Huos
thet
market
var
eth
huss
by
this
monastery.acc
was
sea-nom
and
en
market.nom
was
there
by
By
that
market.acc
was
en
house.nom
âBy this monastery, there was sea and a market was there. By this market, there was a house.â (DA_Mar, Period II)
Otherwise, the demonstrative is never used without an adjective in weak form and (in Swedish) without the clitic -in.
The compound denne is typically found with discourse referents that have been backgrounded for some time and are reinstated as topics.
(157)
Thet
war
en
timæ
en
hellugh
møø
aff
tyræ
ther
hob
hafdæ
til
Vorherræ
At
hun
skulde
syn
mødom
till
tiæniste,
oc
til
æræ
hun
var
ey
en
ælliuffue
vinter
gammell
Hennis
fader
var
aff
høgh
slæct
oc
han
var
ridd-er-s
forman,
oc
mester
Hennis
fader,
oc
hennis
modær
vare
af
enkyns
folk
slæct
ther
ethnici
hetæ
oc
the
hafde
thennæ
iomfruge
til
enugh
dotter
och
hun
hedh
cristina
There
was
en
time
en
holy
maid
of
Tyr
who
hope
had
to
Our.Lord
that
she
should
her.refl
maidenhood
to
service
and
to
glory
she
was
not
yet
eleven
winter
old.
her
father
was
of
high
family
and
he
was
knight-pl-gen
headman
and
master
her
father
and
her
mother
were
of
kind
folk
family
which
heathen
was.called
and
they
had
this
maid
to
only
daughter
and
she
was.called
cristina
âThere was one time a holy maid of Tyr who hoped to Our Lord that she should her maidenhood keep to service and glory, she was not yet eleven winters old. Her father was of high family, he was knightsâ chief and master. Her father and her mother were of a heathen family and they had this maid as their only daughter, she was called Cristina.â (DA_Kerst, Period II)
Otherwise, we also find the compound denne when there is a risk of confusing two discourse referents. In (158) below there are two nuns described, Saint Catharine and another, who is jealous of Catherine and always trying to wrong the saint. When reintroduced in the text, the saint is presented by means of not only a demonstrative but also her full name.
(158)
Ther
thenne
heligh
jomfrw
sancta
katherina
saw
oc
fornam
thet.
tha
arbeydethe
hwn
ther
om
meth
all
øtmyghet
och
alle
the
thieneste
oc
kærlighet-z
gærninghe.
som
hwn
kunne
gøræ
henne.
at
hwn
matthe
mildgøræ
hennes
wredhe
oc
forbannethe
affwyndh
som
hwn
till
henne
haffde
when
this
holy
maid
saint
Catherine
saw
and
understood
that
then
worked
she
on
this
with
all
humility
and
all
the
service
and
love-gen
deeds
which
she
could
make
her
that
she
may
placate
her
fury
and
damned
jealousy
which
she
to
her
had.
âWhen this holy maid Saint Catherine realized this [that the other nun hated her], she worked humbly to overcome her fury and jealousy by service and loving deeds.â (DA_Kat, Period III)
The marking of least accessible referents takes the form of definite descriptions. In Arielâs Accessibility Marking Scale two types of definite descriptions are included: short, with just the definite noun, and long, with some additional modifiers. We retain this distinction, but note that in our material we find that direct anaphors which are definite NPs are almost entirely of the short variety, as illustrated in Table 62.
In absolute numbers, there are more long definite descriptions used as direct anaphors in Swedish than in either Danish or Icelandic, with their highest concentration in Period II. On closer inspection, these long descriptions serve the purpose of differentiating between two discourse referents described by the same noun, e.g., man âmanâ. The long definite description in (159) is used to avoid potential confusion of the competing referents. When there is no such risk the short definite description is used, in this case even despite the long distance between the two mentions of the same discourse referent.
Table 62
Short and long definite descriptions in direct anaphoric contexts
|
Type of |
Short definite descriptions as direct |
Long definite descriptions as direct |
||||
|---|---|---|---|---|---|---|
|
description |
anaphors |
anaphors |
||||
|
Period |
1200â1350 |
1350â1450 |
1450â1550 |
1200â1350 |
1350â1450 |
1450â1550 |
|
Danish |
31 |
83 |
74 |
0 |
4 |
0 |
|
Swedish |
50 |
212 |
46 |
2 |
16 |
11 |
|
Icelandic |
86 |
42 |
47 |
0 |
0 |
0 |
(159)
Thz
war
en
man
som
pläghadhe
hawa
bistokka
j
sinom
gardh.
En
annar
man
kände
hanum
ok
sagdhe
wil-t
thu
at
thin
bij
aldrigh
bort-fliugha
tha
tak
gud-z
likama
ok
läg
j
en
aff
thinom
bistokk-um
Thän
fawizske
mann-in
giordhe
som
hanum
war
känt
ok
gömde
gud-z
likama
j
en-om
stok
There
was
en
man
who
used.to
have
beehives
in
his.refl
garden.
en
other
man
knew
him
and
said
want-2sg
you
that
your
bees.nom
never
away-fly
then
take
god-gen
body
and
place
in
en
of
your
beehive-dat.pl
def
stupid
man-def
did
as
him
was
known
and
hid
god-gen
body
in
en-dat
hive
[12 syntagms without mentioning either of the men]
Tha
thän
tim-in
kom
at
mann-in
wille
sla
sina
stokk-a.
ok
taka
honagh-in
then
this
time-def
came
that
man-def
want
open
his.refl
hive-acc.pl
and
take
honey-def
âThere was a man who used to have beehives in his garden. Another man knew him and said: If you want your bees to never fly away, then take Godâs body and place it in one of your beehives. The stupid man did as him was known and hid Godâs body in one of the hives. [â¦] Then the time came that the man wanted to open his hives and take the honey.â (SV_Jart, Period II)
One reviewer points out the adjective used in the long definite description may also be taken to be an epithet rather than a descriptive adjective, as it is evaluative (presents the speakerâs attitude to the referent rather than describes the referentâs qualities). Epithets are noun phrases that are treated as pronominal elements, i.e., the referent is so topical or salient that a pronoun could be used instead of the epithet, compare the following examples from English:
(160) Hans has called again. I donât want to hear anything anymore from that idiot.
(161) Hans has called again. I donât want to hear anything anymore from him.
In German, the strong definite article can be used. F. Schwarz gives these examples as von dem Idioten âfrom the idiotâ (without contraction of the preposition and the definite article) and gives the contracted form vom Idioten as incorrect (Schwarz F. 2009:31). However, the adjective âstupidâ can also be taken to be descriptive and in the example quoted above its addition clearly serves to differentiate between two referents.
Similarly, in a different Swedish text, there are two discourse referents that can only be told apart by their size, and are therefore referred to with long definite descriptions such as den litla granen âthe little spruceâ and den stora granen âthe large spruceâ (SV_SVM, Period II).
Based on the corpus data we may conclude that long definite descriptions are more likely to be used as direct anaphors when there are competing referents, but not necessarily due to the distance between the two mentions of the same discourse referent.
5.3.1.3 Interim Summary
The proposed Accessibility Marking Scale began with BNs for the most accessible discourse referents, and concluded with long definite descriptions for the least accessible ones.
(162) BN > NP + (adjectival) modifier > NP + demonstrative (+modifier) > short definite description > long definite description
The examples found in the corpus reveal that, while BNs are grammatical in direct anaphoric contexts in Period I in all languages (see also Chapter 4), their continued use as anaphors is only possible under âextenuating circumstancesâ, such as the discourse referent being not only familiar but also unique (as in example 153). Another important limitation on their use is that anaphoric BNs are restricted to neighbouring syntagms.
NPs with adjectival modifiers but without determiners were found sporadically in this context, usually with an anaphoric adjective (âaforementionedâ etc.). There are also very few long definite descriptions amongst direct anaphors.
Among the factors influencing the choice of form, we find that the distance between the antecedent and the anaphor is important for the use of BNs as direct anaphors (if the distance is short, BNs are allowed) or definite descriptions and demonstratives (if the distance is more than one syntagm, either demonstratives or definite descriptions are used). The number of competing referents influences the choice between short (no competitors) and long (some competition) definite descriptions. The topicality of the discourse referent also plays a part, as we will illustrate in the next section.
5.3.1.4 Anaphoric Marking in Anaphoric Chains
We have noted that different NP types could be used in direct anaphoric contexts, the choice between them being partly determined by the age of the text (BNs were preferred in the oldest extant text, but then gradually abandoned in favour of the other NP types) and the accessibility of the discourse referent (according to the proposed scale). We have already presented examples of variation in direct anaphoric marking within one and the same text or even passage, in what are called anaphoric chains. We will now examine these more closely.
Anaphoric chains are sequences of anaphoric links between consecutive anaphors, which all share the same antecedent. It seems more accurate to speak of links rather than individual relations between the anaphors and the antecedent, mainly because the antecedent may change during a sequence of events described in the passage. Consider the following example:
(163) Wash a bunch of fresh spinach well and then shred it finely. Sauté it in a little butter until it is wilted, drain ___ (zero), then put a little into each ramekin.
(Stirling and Huddleston 2002:1457)
We may note that with each action prescribed by the text, the referent spinach is different from what it was before that action took place: before it is sautéed it is shredded, after it is sautéed it wilts and so on. Therefore, we choose to consider such chains as sequences of links rather than collections of anaphors.
The fact that a variety of anaphoric markers may be applied within one anaphoric chain is not surprisingâthe choice of marker depends on a number of factors as discussed above, and these may differ for the same referent within one text passage. In the corpus, however, it is interesting to find that, even in texts which regularly use definite NPs in direct anaphoric contexts, BNs still occur in anaphoric chains. This fact has already been noted in a number of earlier publications (as well as in Chapter 4, section 4.6.1). In view of referent accessibility, we may conclude that there is hardly any difference between the direct anaphors in terms of distance or competing referents (see examples 132â133, section 4.6.1). Below we quote similar examples from three texts in the corpus. Seemingly, Icelandic (example 164) is most liberal in allowing BNs in direct anaphoric contexts alongside definite NPs, but we find similar variation within one passage in both Danish (example 165) and Swedish (example 166), although these are limited to Period I.
(164)
Loðhöttr
hét
þræll
sá,
er þar
var
fyrir stjórn
annar-ra
þræl-a.
Ãessi
þræll
var
kvángaðr,
ok
ól
kona
hans
son
jafnframt
þvÃ,
sem
dróttning
varð
léttari
Ok
þessi
sveinn
var
svá
undarliga
fagr,
sem
þræls-kona-n
átti,
at
dróttning
þóttist
ekki
lýti
sjá
á
svein-i-n-um,
ok
sýndist
henni
nú
þessi
sveinn
ástúðligri
en
sÃnir
svein-ar.
SÃðan
ræðir
dróttning
til
kaups
um
svein-a
við
ambátt-ina.
En
ambátt-inni
sýndist
svá
sem
dróttning-u,
at
henni
þótti
sinn
sonr
tÃguligri,
en
þorði
þó
eigi
at
synja
at
kaupa
við
dróttning-u
um
svein-a-n-a.
Ok
tekr
dróttning
við
ambáttarsyn-i
ok
lætr
nafn
gefa
ok
kallar
svein-inn
Leif.
Loðhöttr
was.called
thrall
that
who
was
foreman
other-gen
thrall-gen.pl
that
thrall
was
married
and
bore
wife
his
son
equally
this
which
queen
was
birth.giving
and
this
boy
was
so
wonderfully
beautiful
which
thrall-wife-def
had
that
queen
could
no
blemish
see
on
boy-dat-def-dat
and
seemed
her
now
this
boy
lovelier
than
her.refl
boy-pl
then
talks
queen
to
swap
about
boy-acc.pl
with
bondswoman-def.acc
but
bondswoman-def.dat
seemed
so
as
queen-dat
that
her
liked
her.refl
son
better
but
dared
though
not
to
refuse
to
swap
with
queen-dat
about
boy-acc.pl-def-acc
and
takes
queen
by
bondswomanson-dat
and
lets
name
give
and
calls
boy-def.acc
Leif
âLoðhöttr was called this thrall who was foreman over the other thralls. This thrall was married, and bore his wife a son at the same time as the queen gave birth. And this boy, that the thrallâs wife had, was so marvellously beautiful that the queen could not see any blemish on the boy. It seemed to her now that this boy was lovelier than her own boys. Then the queen wants to swap the boys with the bondswoman. But to the bondswoman it seemed just like to the queen, that she liked her own son better, but she dared not refuse to swap the boys with the queen. And so takes the queen the son of the bondswoman, and calls the boy Leif.â (IS_Ge, Period I)
(165)
Sak
ængin
skal
af
byhd
stæfna
num
man
skilis
opinbarlika
uiþær
kunu
sinæ
at
hionæ-lah
æn
allæ
andræ
sakær
stande
u-sota
til
biskop
kumbær
i
þe
byhd [â¦]
um
annær
man
uarþær
sæktæþær
ok
skiutær
sik
undæn
ok
uil
æi
biskop-s
i
byhd
biþæ
þa
skal
han
længre
um
land
æftir
ualkæs
mæn
skal
huær
til
loha
takæ
først
af
by
sinum
i
lahfasta
man-na
eþ
suit
none
shall
of
village
sue
unless
man
separates
clearly
with
wife
his
with
marital-law
and
all
other
suits
stand
not-sued
until
bishop
comes
to
that
village
if
other
man
is
accused
and
withdraws
himself
away
and
wants
not
bishop-gen
in
village
beg
then
shall
he
farther
about
land
after
look.for
men
shall
each
to
law
take
first
of
village
their.refl
in
lawful
men-gen.pl
oath
âA village shall not sue unless a man separates clearly from his wife of marital law, and all other suits are pending until a bishop comes to this village. [â¦] If another man is accused and withdraws himself, and does not want to beg the bishop in the village, then he shall be looked for [detained] farther away. Men shall each take to the law, first, men from the village in a lawful menâs oathâ (DA_SKL, Period I)
(166)
En
tima
þær
iohannes
prædicaþe
fik
han
se
en
ungan
suen
vænan
ok
þo
vildan
iohannes
fik
biskop-e
suen-en
ii
hand [â¦]
Iohannes
kom
atar
tel
staþ-en
ok
sporþe
at
suen-en-om
biscop-en
sagþe
hanom
vm
suen-en
som
sant
var
Johannes
gaf
biskop
skuld
for
vangømo
en
time
there
Johannes
preached
got
he
see
en
young
boy
beautiful
and
though
wild
Johannes
got
bishop-dat
boy-def
in
hand
Johannes
came
back
to
town-def
and
asked
about
boy-def-dat
bishop-def
said
him
about
boy-def
which
true
was
Johannes
gave
bishop.dat
guilt
for
negligence
âOne time as Johannes was preaching, he saw a young boy, beautiful but wild. Johannes gave a/the bishop charge of the boy. [â¦] Johannes returned to the town and asked about the boy. The bishop told him the truth. Johannes blamed the bishop for negligence.â (SV_Bur, Period I; quoted after Skrzypek 2012:93â94)
In each of the three examples quoted above there is a BN anaphor in the anaphoric chain. In these passages the common denominator seems to be the topicality of the discourse referent. In the Icelandic example, first the son is put in focus, and referred to with demonstratives and definite NPs. Then the topic shifts to the queen (who, as a unique referent, is presented as a BN) and the mother, while the boy is backgrounded and appears as a BN. In the Danish example, the village is first in focus, as the place of the bishopâs visit, but when it is referred to by means of a BN it is backgrounded in favour of the bishop, who is now in focus. Thus, even though the distance between the previous mention and the anaphor is significant, the discourse referent is a BN. Finally, in the Swedish example the bishop is only presented as a definite NP when it is topicalized (clause-initial subject); in other instances a BN is used.
5.3.1.5 From Deixis to Direct AnaphoraâA Brief Summary
Variation in direct anaphoric marking, which is natural in all languages irrespective of whether or not they have grammaticalized articles, was also present in North Germanic texts written between 1200 and 1550. We have observed, however, that while BNs could be part of anaphoric chains in Period I, they gradually disappear in favour of definite NPs. We also find that the choice of anaphoric marking was determined by factors identified in previous research on referent accessibility, such as distance between the antecedent and the anaphor, the number of competitors for the role of antecedent, and the topicality of the discourse referent.
5.3.2 From Direct Anaphora to Indirect Anaphora
The first stage of definite article grammaticalizationâthe use of the demonstrative as an anaphoric markerâmay be explained as an extension of situational deixis to textual deixis. It does not violate the original meaning of the grammaticalizing form, as its function is still to help the hearer identify a referentâa discourse referentâamong other potential referents. Demonstratives can possibly be used to refer textually in any language, and the fact that they are used in this context does not necessarily mean that the grammaticalization of the demonstrative into an article is ongoing.
The second stage of definite article grammaticalization is the further extension of the original demonstrative to refer to entities newly introduced into the discourse, but connected with some entities or events mentioned earlier. This use is anaphoric and yet indirect, since the anaphor is not co-referential with any other NP in the previous text.
(167) I bought an interesting book yesterday. The author is Nigerian.
The referent of the definite NP âthe authorâ must be retrieved indirectly via the previously mentioned referent âan interesting bookâ. It is generally agreed that demonstratives cannot serve as indirect anaphoric markers (though see Charolles 1990, Kleiber 1990 and Apothéloz and Reichler-Béguelin 1999 for attempts to prove otherwise). De Mulder and Carlier argue that this shift in the use of the demonstrative is possible because the source of the definite article is not merely a demonstrative but a distal demonstrative and thus more vague, and it âcan be understood as an invitation addressed to the hearer to mobilize previous knowledge in order to retrieve the referentâ (De Mulder and Carlier 2011:530). In this way, the distal demonstrative can denote a referent that is not fully identifiable, appealing to specific knowledge shared by the speaker and the hearer (see also Himmelmann 2001:833).
We have already noted that the definite article in North Germanic is derived from a distal demonstrative, which is a marker located low on Arielâs Accessibility Marking Scale. Indirect anaphora is a case of even lower accessibility, since the referent is in fact new, though accessible through a different referent or event. However, we have also noted that, particularly in long anaphoric chains, the original discourse referent introduced at the beginning of the passage need not be (fully) identical to the discourse referent denoted by the last anaphoric expression in the chain (see example 163). Less spectacularly, subsequent mentions of the referent may furnish us with new information regarding its properties. There is a difference between direct anaphors that are (lexically) exact copies of their antecedents (example 168) and those that add new lexical information (example 169).
(168) I came into a room. [â¦] The room was large.
(169) This autumn a new boy started school with us. The quiet newcomer was the subject of gossip for a while.
While we might argue that although ânew boyâ and ânewcomerâ basically share the same semantic content, there is no indication of whether the new boy was quiet or rowdy or had any other qualities that are included in the anaphoric NP. We do not know that the newcomer is a male and so forth. In other words, new information may be incorporated in the co-referring definite NP, while some of the old information may be omitted. Similarly, a different term might be used to refer back to the antecedent, e.g., a bookstoreâthe shop.
In (170) we find an Icelandic example of this type, where the antecedent, kjúklingar âgoslingsâ, is referred to anaphorically by means of the wider term fugla-na âbirds-defâ.
(170)
SÃðan
tók
Grettir
við
heim-gás-unum.
Ãær
voru
fimm
tigir
og
með
kjúkling-ar
margir.
Eigi
leið
langt
áður
honum
þóttu
þær
heldur
bágrækar
en
kjúkling-ar
seinfærir.
Honum
gerði
mjög
hermt
við
þessu
þvÃ
að
hann
var
lÃtill
skapdeildar-maður
Nokkuru
sÃðar
fundu
föru-menn
kjúkling-a
dauða
úti
og
heim-gæs
væng-brotnar.
Ãetta
var
um
haust-ið.
Ãsmundi
lÃkaði
stórilla
og
spurði
hvort
Grettir
hefði
drepið
fugl-a-n-a.
then
took
Grettir
by
home-geese-def.dat.pl
there
were
five
tens
and
with
gosling-pl
many
not
road
far
before
him
seemed
that
rather
unsteady
and
gosling-pl
troublesome
him
made
very
vexed
with
these
for
that
he
was
meagre
temper-master
some.time
later
found
wayfaring-men
gosling-acc.pl
dead
out
and
home-geese
wing-broken
that
was
about
autumn-def
Asmund
liked
badly
and
asked
if
Grettir
has
killed
bird-acc.pl-def-acc
âSo Grettir went to mind the geese. There were fifty of them, and a number of goslings. Before long he began to find them troublesome, and the goslings would not come on quickly enough. This put him out, for he could never control his temper. Soon afterwards some wanderers found the goslings lying outside dead, and the geese with their wings broken. That was in the autumn. Asmund was very much annoyed and asked Grettir whether he had killed the birds.â (IS_Gret, Period I)
In (170) the hearer must apply their wider knowledge to successfully interpret the final definite NP in the anaphoric chain, i.e., they must know that geese and goslings are birds.
However, it is examples such as (171) below that seem to constitute the bridging context between direct and indirect anaphora. We quote the passage in full to give the reader an idea of the whole story, which we also summarize below.
(171)
Han
gik
ena
nat
til
hænne
sæng
ther
hon
soff
mz
barn-it
Oc
myrdhe
sins
brodher-s
barn
oc
hænne
sofwande
stak
han
knifw-in
j
hænna
hand
Ãn
hon
waknadhe
aff
thy
at
blodh-it
fløth
vnder
hænna
sidho
Hon
byriadhe
ropa
swa
at
alt
folk-it
waknadhe
oc
komo
thith
løpande
oc
funno
barn-it
dræpit
oc
knifw-in
blodhoghan
j
hænna
hand
Ther
græt
badhe
fadher
oc
modher
oc
alle
the
ther
waro
Tha
kom
ok
then
ful-e
mordhar-in
løpande
oc
grep
the
ærlik-a
qwinn-o-n-a
j
har-it
he
went
en
night
to
her
bed
where
she
slept
with
child-def
and
murdered
his.refl
brother-gen
child
and
her
sleeping
stuck
he
knife-def
in
her
hand
and
she
woke
of
this
that
blood-def
flowed
under
her
side
she
started
scream
so
that
all
folk-def
woke
and
came
there
running
and
found
child-def
killed
and
knife-def
bloody
in
her
hand
there
cried
both
father
and
mother
and
all
they
there
were
then
came
also
def
vile-wk
murderer-def
running
and
seized
def
honest-wk
woman-acc-def-acc
in
hair-def
âOne night he went to the bed, in which she slept with the child, and murdered his nephew and planted the knife in her hand. And she woke feeling blood flow under her side. She started to scream so that everybody woke and came running and found the child killed and the bloody knife in her hand. Mother, father and all that were there cried. Also the vile murderer came running and grabbed the honest woman by her hair.â (SV_ST, Period II)
In this story a young knight, brother to the emperor, fell in love with a woman who had found sanctuary at the emperorâs court and was employed there to mind the emperorâs child. As she spurned the knightâs advances, he first tried to blacken her character, but as this did not work, he murdered the child in her care, leaving a bloodied knife beside her, so that she would be blamed for the deed. The knight is referred to as either en unger riddare âa young knightâ, pronominally (honom âhimâ, sin âhis.reflâ, han âheâ) or, after he is described as the one who myrdhe âmurderedâ his nephew, as then fule mordharin âthe vile murdererâ. Although all NPs in this anaphoric chain are co-referential (âthe young knightâ is identical with âthe vile murdererâ, the latter being an instance of an unfaithful anaphor, see Lundquist 2007), without the intervening description of the murder we would not be able to place the last definite NP in the anaphoric chain. In other words, the final definite NP âthe vile murdererâ is an indirect anaphor anchored by the verb âmurderedâ, while at the same time it is a direct anaphor, with a different lexical content, of the antecedent âthe young knightâ. That the young knight was vile and a murderer is not just new information added by the speaker, it is actually anchored in the text and can be deduced from the text; an old referent is given a new role. We argue that this type of context is a bridging context, allowing the grammaticalizing definite to spread to indirect anaphoric contexts. In the following we will investigate the indirect anaphora, its typology, and the NPs used in this context in the corpus.
5.3.2.1 Indirect AnaphorsâTypology
In 2.2.1 we presented a typology of indirect anaphors, noting that authors differ in their views on bridging reference, although, most of them distinguish between at least two major types of indirect anaphors: mereological indirect anaphors (the anchor is an established discourse referent; this includes partâwhole relations and other thematic roles, e.g., bookâcover, bookâauthor) and frame-related indirect anaphors (the anchor is an eventuality or a frame present in the discourse model). The first major work on the subject, Clark and Haviland (1977), distinguishes between three types of indirect anaphors: meronymic relations, thematic relations and reasons/causes/consequences together with other relations between events. Schwarzâs classification, based on her study of German texts (Schwarz M. 2000), largely follows this tripartite typology and includes four types: meronymy (partâwhole relations), lexical/thematic type (thematic roles), scheme-based conceptual type, and inference-based conceptual type (complex inferencing). Irmer (2011) considers only meronymic and thematic types to be instances of bridging, while inference-based types are in his account excluded from the bridging types. Other notable studies of the context include Fraurud (1986), Gundel, Hedberg and Zacharski (1993), Cornish (1996, 1999), Poesio and Vieira (1998), Matsui (2000), Epstein (2002), Vieira and Poesio (2000), Poesio (2003), Löbner (2003), and an overview in Zhao (2014).
In the present study we largely follow M. Schwarz (2000) (a study based on actual language use and not constructed examples) and a simplified version of her typology in which we distinguish three major types of indirect anaphors: meronymic relations, thematic types (under which we subsume M. Schwarzâs lexical/thematic type and scheme-based conceptual type) and inference-based conceptual type.
Meronymic relations are usually taken to be the canonical example of bridging in most studies. Gardent et al. (2003) demonstrate that partâwhole relations indeed constitute a majority of all instances of bridging (in a study limited to cases with nominal and verbal antecedents). In 2.2.1.2 we presented a typology of meronymic types, consisting of necessary parts, probable parts and inducible parts (see examples 28â36, Chapter 2, section 2.2.1.2).
A typical feature of meronymic relations is that the antecedent is nominal. In many cases it is also natural to substitute a possessive NP for a definite NP, something which will be of consequence in our later analysis. The variation between defNP and possNP can be partly dependent on other factors, such as the animacy of the anchor. This applies especially to so-called inalienable possession, where the indirect anaphor constitutes a part of or is in close spatial relation with a whole that is a living body. We may observe that there are indeed differences between closely related article languages such as English and Swedish with respect to the marking of the indirect anaphor in this type of context (see also Lødrup 2009, 2010, 2014). While English utilizes possessive pronouns, Swedish (like other North Germanic languages) uses the definite article, as in (172).
Swedish
(172)
Jan
stoppade
hand-en
i
ficka-n.
Jan
put
hand-def
in
pocket-def
âJan put his hand in his pocket.â
English
(173) John put his hand in his pocket.
As regards thematic types, for Clark and Haviland (1977) and M. Schwarz (2000) this group consists of semantic roles, such as agent, object and instrument. In Clark and Haviland (1977) they are further subdivided into necessary and optional, such as the relation between to murderâthe murderer (each murder has a murderer) and to murderâthe knife (the instrument of a murder need not be a knife). As these examples indicate, the antecedent need no longer be nominal. Within this group we find different relations between anchor and anaphor, such as objectâmaterial (bicycleâthe steel), objectâcomponent (jokeâthe punchline), collectiveâmember (deckâthe card), massâportion (pieâthe slice), etc. This group of indirect anaphors is less likely to be expressed by possessives.
Here we touch upon one of the puzzles which we presented in Chapter 1, i.e., the puzzle of successâwhy do demonstratives grammaticalize into definite articles but not any other potential candidates, such as possessive pronouns? The typology of indirect anaphora presented above, though very general, nevertheless reveals that the possessives are a viable alternative for a definite article only in a limited number of contexts, where nominal anchor is available, since it is not possible to use the possessives when the anchor is not a nominal. In order to be used in other types of indirect anaphora and, successively, in other definite contexts, the possessives would need to be bleached of the core of their meaning, i.e., the expression of possession, a link with another nominal. To our knowledge, such a process has not been observed in any language. The grammaticalization of the definite article out of a demonstrative, on the other hand, requires only the loss of proximal-distal distinctions, a process which is well-attested in a number of languages, and which seems to occur independent of the grammaticalization, as it is a natural development that such distinctions became blurred or lost with time.
Clark and Haviland distinguish a third group, which comprises types such as reasons, causes, consequences and other relations between events. M. Schwarzâs third group is called inferential types, and comprises anaphors whose resolution builds on world knowledge and cannot be derived from the discourse alone. This group is not regarded as a type of bridging by Irmer (2011), who instead classifies it as a coherence device (see also Givón 1992 and 1995).
(174) Wussten Sie [â¦] dass der Schrei in Hitchcocks âPsychoâ deshalb so echt wirkt, weil der Regisseur genau in dem Moment der Aufnahme eiskaltes Wasser durch die Leitung pumpen lieÃ?
âDid you know [â¦] that the scream in Hitchcockâs Psycho seems so real because at the moment of filming the director allowed cold water to be pumped through the pipe?â
(Consten 2004:102)
To successfully interpret an anaphor of the inference-based conceptual type, some general knowledge is necessary. The interpretation of the definite NP die Leitung âthe pipeâ relies on familiarity with the Hitchcock film and the fact that the famous scene with the scream takes place in a shower. In this context the indirect anaphor can only be expressed by a definite NP (in an article language), while demonstratives and possessives are disallowed in this context.
Traditionally, the point of departure for all classifications has been linked to definite NPs without an antecedent. Studies have aimed to explain their definiteness in the absence of an antecedent. However, in recent years, when the concept of bridging has become more established, more and more authors appreciate that bridging can occur also in the absence of definites (e.g., Asher and Lascarides 1998:107). In his discussion of totality (i.e., exhaustivity, completeness), Hawkins (1978) demonstrates that the definite can only occur in bridging when the referent is unique, e.g., a carâthe engine but a carâa tyre, yet the underlying relationship between the engine and the car seems to be the same as that between a tyre and the car. It has also been demonstrated that possessives may introduce new, yet anchored referents (Willemse, Davidse and Heyvaert 2009). Those authors found that in a considerable number of cases the PM (= possessum) referents of possessive NPs are first mentions with inferential relations to the context (Willemse, Davidse and Heyvaert 2009:24), just as many definite NPs introduce new, grounded referents. It has in fact been noted that some contexts allow choice between defNP and possNP, as in example (175) (after Fraurud 2001:246).
(175) Beside the barn there is a little cottage. The/Its roof is leaking.
(but: *This roof is leaking)
There is, however, at least one obvious difference between bridging with definites and bridging with possessives: the definite article makes an implicit link with the other nominal, while the possessive makes an explicit link. Possessive NPs need a nominal antecedent, while definite NPs can be anchored by verbs, clauses or the text itself.
Bridging can be defined more or less broadly, depending on the types of relationships, the types of antecedents and the types of anaphors. For the present study we have chosen to adopt the following model of bridging relations:
-
the anaphor can be any nominal NP: BN, defNP, indefNP, possNP or NP with a different determiner;
-
the antecedent may be nominal or verbal, but the anaphor may also be resolved within the common ground (in the sense of Stalnaker 2002), thus lacking an explicit textual anchor.
We classify all indirect anaphors in the following groups:
-
meronymic relations
-
thematic types
-
inference-based conceptual types
We are also fully aware of the imperfections of the model and the crudeness of this typology. However, our goal is first and foremost to find the types of bridges that adopt the incipient definite article early, and those that adopt it later or perhaps express the bridging relation by other means. Only then will we be able to refine our classification.
Considering that there are so many subtypes of indirect anaphora, the next question to be asked is: do they all allow the incipient definite article at the same time, or do any of the types appear as definite NPs earlier? In other words, is there a greater affinity between one or more types of indirect anaphora and direct anaphora? In the remainder of this section, we will consider all three types of indirect anaphors separately, beginning with meronymic relations.
5.3.2.2 Semantic Types: Meronymic Relations
A number of examples of meronymic relations are found in the corpus. However, with limited material at our disposal, we were not able to find examples of each type of meronymic relation, which would enable a systematic study of all subtypes for all periods, and which would allow us to account for the possible differences of expression between necessary, probable and inducible parts. The examples which are best represented are those of inalienable possession, i.e., indirect anaphors referring to body-parts, items of clothing or weaponry, and anchors denoting human referents.
The types of NPs found in meronymic relations include bare nouns, possessive NPs and definite NPs, although in Period I inalienables seem to be found only as BNs (example 176) or possNPs (examples 177â178) and not as defNPs.
(176)
iak
kom
þa
fuul
sørhilika
til
miin
kæra
sun
ok
þahar
iak
sa
hanum
slaa-s
mæþ
næua [â¦]
ok
spytta-s
i
anlæt
ok
krona-s
mæþ
þorna
I
came
then
fully
sorrowful
to
my
dear
son
and
when
I
saw
him
beat-refl
with
fists
and
spit-refl
in
face
and
crown-refl
with
thorns
âI came full of sorrow to my dear son and as I saw he was beaten with fists [â¦] and spat in the face and crowned with thornsâ (DA_Mar, Period II)
(177)
Ãn
of
swa
worthær
at
man
mistær
allæ
sinæ
tændær
af
sin
høs [â¦]
if
of
so
becomes
that
man
loses
all
his.refl
teeth
of
his.refl
head
âIf so happens that a man loses all his teeth from his head [â¦]â (DA_VL, Period I)
(178)
at
hon
varþ
hauande
mæþ
gu-z
son
ii
sino
liue
that
she
became
pregnant
with
God-gen
son
in
her.refl
womb
âThat she became pregnant with Godâs son in her womb.â (SV_Bur, Period I)
In Period II inalienables no longer appear as BNs, but with either a (reflexive) possessive pronoun or the incipient definite article. It should be noted here that North Germanic languages have retained two possessive pronouns: the regular possessive hans/hennes/dess, corresponding to the English his/her/its, and the reflexive possessive, sin/sitt, which is used when the possessor is the subject of the clause. The default marking of inalienables in Period II seems to be the reflexive possessive, as in (179), and the incipient definite article is at first found with inalienables only in direct anaphora (i.e., such body-parts or items of clothing that are not only connected with an owner known from previous discourse, but have also been mentioned themselves).
(179)
Kwinna-n
gik
bort
ok
faldadhe
han
j
sinom
hwiff
som
hon
hafdhe
a
sino
hofdhe
woman-def
went
away
and
folded
him
in
her.refl
scarf
which
she
had
on
her.refl
head
âThe woman went away and folded him in her scarf, which she had on her headâ. (SV_Järt, Period II)
(180)
Tha
synti-s
quinn-onna
hwifwir
allir
blodhoghir
ok
water
aff
blodh
swa
at
blodh-in
flöt
nidhir
vm
quinn-onna
kindir.
Hulkit
herra-n
saa,
ropadhe
ok.
sagdhe
hwar
slo
thik
j
thit
änlite
älla
sarghadhe.
Ok
quinnna-n
lypte
vp
sina
hand
ok
strök
sik
vm
änlit-it
ok
tha
hon
tok
nidhir
hand-in-a
tha
war
hon
al
blodhogh.
then
see-refl
woman-def.gen
scarf
all
bloody
and
wet
of
blood
so
that
blood-def
flowed
down
about
woman-def.gen
cheeks
which
master-def
saw
screamed
and
said
who
hit
you
in
your
face
or
hurt
and
woman-def
lifted
up
her.refl
hand
and
stroked
herself
about
face-def
and
when
she
took
down
hand-def-acc
then
was
she
all
bloody
âThen the womanâs scarf seemed all bloodied and wet with blood so that the blood flowed down the womanâs cheeks. Which the master saw, screamed and said âWho hit you in your face or hurt you?â And the woman lifted her hand and stroked her face and when she took the hand away it was all bloodied.â (SV_Järt, Period II)
Example (180) illustrates well the division of labour between the (reflexive) possessive and the incipient definite article. The possessive is used if the inalienable possessum is mentioned for the first time (indirect anaphora). The definite article is used only in further mentions, i.e., in direct anaphora (thus your faceâthe face, her handâthe hand). Naturally, we could simply treat such examples as direct anaphors. However, it is clear that they are both co-referential with an antecedent and accessible via their anchors. It seems that this double identity, as direct and indirect anaphors, constitutes a bridging context (in the sense of Heine 2002) for definite NPs to spread to indirect anaphora with meronyms. By the end of Period II and the beginning of Period III, the definite article begins also to be used in indirect anaphora (first mention of an inalienable possessum connected with a known discourse referent), as presented in (181) and (182).
(181)
Tha
bar
keysar-in
vp
hand-en-a
oc
slogh
hona
widh
kinben-it
at
hon
størte
til
iordh-inna
then
bore
emperor-def
up
hand-def-acc
and
hit
her
at
cheekbone-def
that
she
fell
to
earth-def
âThen the emperor lifted his hand and hit her on the cheekbone so that she fell down.â (SV_ST, Period II)
(182)
Tha
begynthe
løffw-en
som
hwn
war
wan
gladeligh
at
løpe
i
closter-eth [â¦]
eller
rørdhe
stiærth-en
then
began
lion-def
as
she
was
used.to
gladly
to
run
in
monastery-def
or
wagged
tail-def
âThen the lion began, as she was accustomed to, to gladly run in the monastery [â¦] or wagged her tail.â (DA_Jer, Period III)
It should be noted that BNs are found in indirect anaphora even in Period III; however, as illustrated in examples (183) and (184), these occurrences may be lexicalizations rather than indirect anaphors.
(183)
Jamwnd-z
hoffuit
bløde
bodhe
giømmen
mwn
ok
øren
Jamund-gen
head
bled
both
through
mouth
and
ears
âJamundâs head bled both through mouth and ears.â (DA_KM, Period III)
(184)
badh
meth
mwndh
oc
hiærthe
prayed
with
mouth
and
heart
â[She] prayed with mouth and heart.â (DA_Kat, Period III)
5.3.2.3 Thematic Types
This type of indirect anaphora is based on our lexical knowledge of certain elements forming more or less stereotypical events or processes, and on our familiarity with stereotypical relations between objects or events and objects; for instance, a court case involves a judge, one or more hearings, a charge, a plaintiff, and so on. Thus, whenever such an event is presented in a text, the hearer will treat definite NPs such as the judge, the hearings, etc. as connected with that event.
In Period I we find mostly BNs in this type of indirect anaphora (examples 185 and 186), but, interestingly, a few instances of the incipient definite article occur as well (example 187).
(185)
Uærþær
maþer
dræpin [â¦]
þa
skal
uighi
a
þingi
lysæ
be
man
killed
then
shall
murder
on
thing
declare
âIf a man is killed [â¦] then the murder shall be made public on a thingâ (SV_AVL, Period I)
(186)
Sitær
konæ
i
bo
dör
bonde [â¦]
sits
wife
in
house
dies
husband
âIf a wife is alive and the husband dies [â¦].â (SV_AVL, Period I)
(187)
Nu
dræpær
maþær
man
koma
til
arua
man-zs-in-s
ok
fa
drapar-an
ok
hugga
þær
niþær
a
fötær
þæs
döþ-a [â¦]
now
kills
man.nom
man.acc
come
to
heir
man-gen-def-gen
and
get
killer-def.acc
and
cut
there
down
on
feet
this.gen
dead-gen
âIf a man kills another, the heirs of the deceased come and get the killer and cut him down at the feet of the deceased [â¦]â (SV_OgL, Period I)
This context allows definite NPs as early as Period I. In Period II the thematic types are regularly found as defNPs, in pairs such as tjuven âthe thiefââstölden âthe larcenyâ, wighia âto ordainââvixlenne âthe ordinationâ, henger âhangsââgalghan âthe gallowsâ, rida âto rideââhästen âthe horseâ, fördes död âa dead (man) was carriedââbaren âthe stretcherâ. A characteristic of this type of indirect anaphora is that the anchor need not be nominal and the anaphor may be accessible through a VP.
Possessive NPs are seldom found in this type at all, independent of the period. We have located some examples of possNPs that may be considered indirect anaphors; it should be noted that, as in example (188), they sound natural with a reflexive possessive in Modern Swedish as well, and the choice between defNP and possNP may be a question of stylistics rather than grammatical correctness.
(188)
Diäfwl-en
saa
hans
dirue
oc
reede
hanom
snaru. [â¦]
Oc
baþ
munk-in
sik
inläta
i
sin
cella.
devil-def
saw
his
courage
and
prepared
him
trap
and
asked
monk-def
himself
allow
in
his.refl
cell
âThe devil saw his courage and prepared a trap for him. [â¦] And he [the devil] asked the monk to let him in his [= the monkâs] cell.â (SV_HML, Period II)
(189)
Ther
sancta
katherina
thette
fornam
tha
luckthe
hwn
sik
hardeligh
i
syn
cellæ
och
badh
jnderligh
till
gudh
when
saint
Catherine
this
understood
then
locked
she
herself
firmly
in
her.refl
cell
and
prayed
passionately
to
God
âWhen saint Catherine understood this, she then locked herself firmly in her [own] cell and prayed passionately to God.â (DA_Kat, Period III)
5.3.2.4 Inference-Based Conceptual Types
This type of indirect anaphora is the least accessible. To correctly identify the referent, the hearer must not only consider textual information or stereotypical knowledge of the world, but also make inferences allowing them to resolve the anaphor. It should be noted that some authors do not consider this type anaphoric at all, e.g., Irmer (2011).
In the corpus, this type is expressed either by bare nouns or by definite NPs. No possessive NPs were found here. An interesting fact, though, is that defNPs may be found as early as Period I, as in (190).
(190)
Maþær
far
sær
aþalkono
gætær
uiþ
barn
dör
sv
fær
aþra
gætær
viþ
barn
far
hina
þriðiu
þör
bonde
þa
konæ
er
livændi
þa
skal
af
takæ
hemfylgh
sinæ
alt
þet
ær
vnöt
ær
hun
ællær
hænær
börn
þa
skal
hin
ælsti
koldær
boskipti
kræfiæ
takær
af
þriþiung
af
bo-no.
man
gets
himself
wife
begets
by
child
dies
this
gets
another
gets
by
child
gets
that
third
dies
peasant
then
wife
is
alive
then
shall
of
take
dowry
her.refl
all
that
which
unused
is
she
or
her
children
then
shall
that
oldest
brood
division
demand
take
of
third
of
property-def.dat
âA man gets himself a wife, she gets with child, then dies, he gets another, she gets with child, he gets a third wife and he dies. Then if a wife is alive, she shall take all her dowry which is not spent. If she or her children are alive, then the oldest brood shall demand division and take a third part of the estate.â (SV_AVL, Period I)
(191)
Nu
j
the
stund-inne
for
ther
fram vm
en
prästir
mz
gud-z
likama
til
en
siuk-an
man
ok
klokka-n
ringde
for
gud-z
likama.
now
in
this
hour-def.dat
travelled
there
forward
en
priest
with
god-gen
body
to
en
sick-acc
man
and
bell-def
rang
for
God-gen
body
âAt this hour a priest was travelling to a sick man, carrying the wafer and the bell rang to announce him.â (SV_Järt, Period II)
We have not found a single example of indirect anaphora that could be classified as a conceptual inference-based type being expressed by a possessive NP. In this type of anaphora defNPs occur earlyâthey are found, though only sporadically, at the beginning of Period I, while, for instance, the meronymic type is not expressed by defNPs until the end of Period II. To begin with, however, BNs are prevalent. Gradually, they are replaced by definite NPs, without going through the possessive NP phase as the meronymic types seem to have done. This type of indirect anaphora may be considered to be reserved for the definite article, since no other element, possessive or demonstrative, can appear here.
5.3.2.5 Summary
Once the incipient definite article begins to occur in indirect anaphoric contexts, it is initially found with a group of contexts which is here referred to as thematic types. This observation ties in well with the type of direct anaphors which we identified as the bridging context between direct and indirect anaphora (example 171). These anaphors are direct in the sense that they have a formal antecedent in previous text, but also indirect in that they identify the discourse referent in a thematic role connected with some event mentioned earlier in the text.
The inferential types are infrequent, but from what we can observe, this context is also relatively early in allowing the incipient definite article. The type of indirect anaphora which lingers longest is the meronymic relation, i.e., the partâwhole relation. Depending on the animacy of the possessor, they appear as possessive NPs (with animate possessors) or BNs (with inanimate entities). Eventually, the entire context adopts the incipient definite article.
5.3.3 From Indirect Anaphora to Larger Situation Use
The larger situation use of the definite article is based not on the familiarity, but the uniqueness of the discourse referent. There is nothing in the preceding discourse which may be identified as an anchor for the definite NP, thus making this definite article use different from indirect anaphora. To identify the referent, the hearer/reader needs to rely on their general knowledge.1
The unique referents that we have found in our corpus come from different domains. The major ones, with most frequently used referents, are nature (sun, earth, air, nature, stars, etc.), religion (God, church, Bible, devil, heaven, hell, faith, etc.), and law and rule (law, king, mayor, emperor, etc.). The majority of these belong to Hawkinsâ (1978, 1991) category of larger situation use, which relies on general (and not specific) knowledge. They may also be considered to be âglobalâ uniques, unvarying for all people and with their uniqueness being absolute (Lyons C. 1999:8).
The only unique reference NP that has survived as a bare noun until modern times in all North Germanic languages is gud âGodâ (used solely with reference to the Christian God, other divine beings may be referred to as gud-ar-na âgod-pl-defâ). Others have gradually adopted the definite article. We can, however, identify certain patterns.
First of all, in the domain âreligionâ some referents are definite as early as in Period I; we find forms such as kerki-an âchurch-nom.defâ or kirki-unna âchurch-acc.defâ in Swedish, and døth-æn-s âdeath-def-genâ in Danish. In Period II we find a number of parallel forms, e.g., himmell-en âheaven-defâ vs. himmel âheavenâ in Danish. As late as in Period III, a number of unique referents from this domain can still occasionally be found as BNs, e.g., and âspiritâ and skriffth âscriptâ in both Danish and Swedish, but it should be noted that they most often appear in definite NPs with an adjectival modifier and a definite preposed article, as in example (192).
(192)
tror
tw
fadher
tha
tror
tw
sønn-æn
ok
then
hælg-e
and
believe
you
father
then
believe
you
son-def
and
def
holy-wk
spirit
âIf you believe in father, then you also believe in the son and the holy spirit.â (Da_KM, Period III)
In the domain ânatureâ some referents lag behind in adopting the definite article (for instance, bare nouns such as sol âsunâ in Swedish or iord âearthâ in Danish are frequent in Period II).
(193)
Jach
hauer
i
thennæ
vkæ
offræt
sannende
offer
til
then
gudh
ther
bodæ
skapeth
himmel
oc
iordh
oc
stiærnæ-r
och
haf
oc
alt
thet
ther
fødh
ær
i
iørde-rige
I
have
in
this
week
offered
true
offer
to
this
god
that
both
created
heaven
and
earth
and
star-pl
and
sea
and
all
that
which
born
is
in
earth-kingdom
âI have this week made true sacrifice to the god who created both heaven and earth and stars and the sea and all that is born on the earth.â (DA_Kerst, Period II)
Other referents, such as wærld-æn âworld-defâ, wæruld-enna âworld-def.genâ, i.e., âthe worldâ, are definite already in the oldest Danish and Swedish texts from Period I. There are also a number of referents in this domain which adopt the definite article relatively early (Period II), and which may be considered either uniques or indirect anaphors, such as skog âforestâ, öken âwilderness, desertâ, hav âseaâ. Examples of such referents are found in (194) and (195).
(194)
Tha
meth
syne
brødh-re-s
raadh
fick
han
løffw-en
thet
æmbeth
at
hwn
skulle
een
asen
som
hænthe
them
weth
aff
skoffw-en
følghe
till
marck-en
ther
han
thog
syn
fødhe
oc
thaghe
hannum
till
waræ
then
with
his.refl
brother-pl-gen
advice
got
he
lion-def
that
task
that
she
should
en
donkey
which
brought
them
wood
of
forest-def
follow
to
pasture-def
where
he
took
his.refl
food
and
take
him
to
care
âThen, by the advice of his brothers, he gave the lioness the task to guard the donkey which brought them wood from the forest and follow him to the pasture where he grazed.â (DA_Jer, Period III)
(195)
I
rom
var
en
keysare.
han
hafdhe
ena
dygdheliga
oc
gudhliga
keysarinne.
hon
thiænte
jomfru
marie
af
allo
hiærta.
keysar-en
vilde
fara
ofvir
mær-it
til
the
hælgh-e
graff.
in
Rome
was
en
emperor
he
had
en
virtuous
and
godly
empress
she
served
virgin
Mary
of
all
heart
emperor-def
wanted
travel
over
sea-def
to
def
holy-wk
tomb
âThere was an emperor in Rome who had a virtuous and pious wife who served Virgin Mary with all her heart. The emperor wanted to travel to the Holy Tomb over the sea.â (DA_ST, Period II)
The reason for treating them as potential indirect anaphors is that they usually appear in connection with some location (such as a city in which the narrative is taking place), and they could be understood as anchored by that location, with the definite article roughly corresponding to âin the vicinity of the cityâ. On the other hand, it is not impossible to consider that at least some of them, such as the sea, may have been considered absolute uniques (like the sun). At the time when the texts were written, users of the language may not have been aware of there being more than one sea in the universe. What separated them from other uniques is also their affinity with the so-called âdefinites of pathâ (see e.g., Roberts 2003), such as âthe elevatorâ in âtake the elevatorâ, when there is more than one to choose from, and yet the NP is definite.
Finally, for the domain ârule and lawâ, we studied a context in which the difference between more local and more global unique referents and their formal representations may be studied diachronically. The context is co-ordinated NPs of the type âto the king, the bishop and the districtâ, which are regularly found in legal prose, stipulating to whom taxes or fines were to be paid. Since the payments were typically divided three ways, such co-ordinated NPs bring together a number of referents which are âdifferentlyâ unique. Furthermore, these NPs have a highly formulaic character and can therefore be expected to retain the original, article-less forms longer. The aim of this study was therefore to establish whether the definite article could be said to appear earlier with some types of referents than with other types. We assumed that the more âlocalâ uniques, such as herad âdistrictâ, socken âparishâ or prest âpriestâ, would be definite earlier than the more âglobalâ ones, such as kung âkingâ or biscop âbishopâ, considering that the former rely on familiarity as well as uniqueness.
This assumption was only partly borne out. True enough, in the oldest extant texts, the unique referents tend to be unmarked and appear as BNs, irrespective of the scope of their uniqueness, as in (196).
(196)
Uerdher
kyrkia
brut-in
oc
mæssu
fat
stol-en [â¦]
þat
er
niv
march-a
sak
kyrky
swa
hæreþe
sva
konogge.
is
church
break-ptcp
and
Mass
plate
steal-ptcp
this
is
nine
mark-pl
fine
church
so
district
so
king
âIf a church is broken into and the Mass plate stolen [â¦]. It is nine marks fine, to church, and so to district and so to king.â (SV_AVL, Period I)
The âglobalâ unique kong âkingâ remains bare in the legal texts Ãldre Västgötalagen, Yngre Västgötalagen and Upplandslagen (1225, 1280 and 1380 respectively), it is sporadically definite in Ãstgötalagen and Dalalagen (both ca. 1280), and the proportions do not differ for other legal texts. Overall, out of all instances of the lexeme âkingâ in the legal texts studied here, 92.2â¯% are bare NPs, while the remaining 8.8â¯% are definite NPs. The first text in which a definite NP âthe kingâ is to be found in about 30â¯% of the instances of that lexeme is Kristoffers Landslag (1440), and this is the highest score.
On the other hand, the more âlocalâ unique höfding âchieftainâ is also predominantly used as a bare noun throughout the legal prose, as in (197), even though instances of its use as a definite NP can be found as early as 1280 (Yngre Västgötalagen), as presented in example (198).
(197)
vil
han
eig
þa
ræt
göræ
þa
nempni
hæræz-höfþingi
ting
wants
he
not
then
right
do
then
calls
district-chieftain
thing
âIf he wonât do what the law says, then the district chieftain calls a thing [assembly].â (SV_YVL, Period I)
(198)
kan
man
hittæ
skiælikæn
ældæræ
kalli
til
sinæ
grannæ
oc
taki
han
oc
före
til
hæræz-höþingi-æn-s
æller
þes
i
hans
staþ
er
satt-er.
can
man
find
true
vagabond
calls
to
his.refl
neighbour.pl
and
take
he
and
lead
to
district-chieftain-def-gen
or
this.gen
in
his
place
is
sit-ptcp
âIf a man finds a true vagabond, he calls to his neighbours and they take the vagabond to the district chieftain or the one who acts in his stead.â (SV_YVL, Period I)
There is also a significant difference in the frequencies of both nouns. While kung âkingâ appears in each legal text between ten and seventy times, höfding âchieftainâ appears sporadically in only four out of eighteen texts. This might partly explain why kung âkingâ remains a bare noun for such a long time. However, all in all, it has to be concluded that the legal texts exhibit no clear pattern in the marking of these unique referents. It is only those referents that could well be considered cases of indirect anaphora that appear as definite NPs earlier, while other referents retain bare forms as late as the mid-15th century, as presented in example (199).
(199)
waren
tha
arfua-ne
saklöse
baade
for
malsegende-n-om,
konung-e
oc
hærade-n-o [â¦]
are
then
heir-def.pl
innocent
both
for
plaintiff-def-dat
king-dat
and
district-def-dat
âIf the heirs are then innocent both according to the plaintiff and king and the district [â¦]â (SV_Kris, Period II)
An interesting exception is the absolute unique pÃ¥ven âthe Popeâ, which appears as a definite NP already in one of the oldest extant legal texts (1225), as in (200).
(200)
All-um
slik-um
mal-um
skal
af
land-i
skiptæ
mæþ
brev-i
til
pavæ-n-s
i
rom
þer
skulu
af
pava-n-um
bref
till
takæ
all-dat
such-dat
case-dat
shall
of
country-dat
change
with
letter-dat
to
pope-def-gen
in
Rome
there
shall
of
pope-def-dat
letter
to
take
âAll such cases of the country shall be changed with a letter to the Pope in Rome, there should a letter be taken from the Pope.â (SV_AVL, Period I)
For a statistical analysis of the larger situation use of the definite article in North Germanic the reader is directed to Piotrowska and Skrzypek (2020).
In conclusion, even though referents connected to the domain of nature are the least frequent among all unique referents in the corpus, they exhibit the largest proportion of definite forms. If we exclude the continuously bare referent gud âGodâ from the results, it is the domain of law and rule that remains unmarked (bare) the longest, while the domain of nature is the first to demonstrate mostly definite referents. Based on these findings we may therefore state that within the larger situation uses, the more local unique referents (i.e., those based on specific knowledge), which could have been interpreted as indirect anaphors and thus textually anchored, seem to have served as a bridge from textual to non-textual definiteness in the grammaticalization of the definite article. The culmination of the grammaticalization is the use of the definite article with globally unique referents, a usage that is at odds with the original meaning of the demonstrative delimiting the referent from other potential referents.
5.3.4 SummaryâA Review of the Grammaticalization Chain of the Definite Article
The data from the North Germanic corpus demonstrate how the successive stages of the grammaticalization chain of the definite article were reached. The original demonstrative was first used to mark direct anaphoric referents, especially those that were topical or reintroduced in the discourse. We find that the spread from direct to indirect anaphora was possible on the basis of contexts which could be read as either type, i.e., there was an antecedent for the anaphor, but the anaphor included new lexical information about the discourse referent. That information identified the referent as a thematic role for some event described earlier in the text.
The use of the incipient definite article in indirect anaphoric contexts marks the turning point in the development of the definite; from that point onwards, the demonstrative and the definite need to be studied separately. While the first stage of the grammaticalization process involved a shift from situational to textual deixis, this stage involves a shift from textual information to information that the speaker believes the hearer to possess and be able to activate in order to identify the referent of the definite NP, although the hearer is still guided by the text.
In Chapter 2, we mention Old French data and Carlier and Simonenkoâs hypothesis on the evolution of the definite article from expressing strong to weak definite semantics. In the French material the authors found that a critical context for the spread from direct to indirect anaphora may be the use of the incipient definite article followed by a restrictive relative clause. In our material, we have identified too few such instances to reinforce this insight; we would like to note, however, that there are no such uses in Period I. A handful of examples were found in Period II in Danish and in Swedish (and none in Icelandic). One Swedish and Danish example is actually identical, as it comes from two translations of the same text; see examples (201) and (202).
(201)
tak
the
yrte-na
som
stondir
vndir
thit
hovith
take
def
herb-def
that
stands
under
your
head
âTake the herb that is under your head.â (DA_ST, Period II)
(202)
Tak
the
yrte-na
som
vnder
thit
hofwdh
staar
take
def
herb-def
which
under
your
head
stands
âTake the herb that is under your head.â (SV_ST, Period II)
Two more Swedish examples were quoted above as (140) and (141).
Although we find examples discussed in Carlier and Simonenko, their frequency is low and their appearance later than the regular use of the definite article in indirect anaphoric contexts, which does not preclude that they did play a role in the evolution of definite article use from direct to indirect anaphora.
The final stage of the development discussed above involves a further shift away from the text and towards the general knowledge the hearer possesses and shares with the speaker. The data presented in the preceding section suggest that such a shift was again possible in contexts with potential double interpretation: on the one hand, as indirect anaphors (e.g., a town is discussed in the preceding discourse, the definite form of mayor may be taken to mean the mayor of that town), and on the other hand, as unique, text-independent reference (âlocallyâ there is only one mayor, and no mention of the town is necessary for the identification of that referent).
We have not discussed the use of definite NPs in generic contexts. Such use marks the final stages of definite (and coincidentally, also indefinite) article grammaticalization. In other words, the articles must be well grammaticalized in other functions to be used generically. In the corpus, we find occasional examples of definite articles in generic contexts, as in (203). However, it is far more common to find BNs used generically as late as in Period III in all languages studied, as in (204) and (205).
(203)
mærk
pa
en
harpæ
ther
ær
træth
strenge
ok
hand-en
ære
iij
tingh
ok
eth
lydh
so
ær
ok
mandell
nøt
then
først
ær
skal-en
ok
swa
fnas-et
ok
so
kærn-in
theære
iij
tingh
ok
eth
note
on
en
harp
there
is
wood
strings
and
hand-def
are
three
thing
and
one
sound
so
is
also
almond
nut
then
first
is
shell-def
and
so
skin-def
and
so
seed-def
there
three
thing
and
one
âNote that as there are wood, strings and handle on a harp but one sound, and in an almond first there is the shell, then the skin, and then the seed, there are three things and one [entity].â (DA_KM, Period III)
(204)
Wen
jac
haffwer
atwakteliga
hört
tyn
ordh,
oc
the
gaffw-or
thyn
konwng
haffwer
oss
sänth
epter
frome
herr-a-s
sedh
when
I
have
carefully
heard
your
words
and
def
gift-pl
your
king
has
us
sent
after
mighty
lord-pl-gen
custom
âWhen I have carefully listened to your words and the gifts that your king has sent us in accordance with the customs of mighty lords.â (SV_Troj, Period III)
(205)
Roland
sade
madkæ
fyskæ
fwglæ
by
worde
ok
leffue
aff
søl-æn
ok
ander
tingh
wdhen
mand-z
samlende
Roland
said
worms
fish
birds
bees
be
and
live
off
sun-def
and
other
thing
without
man-gen
union
âRoland said worms, fish, birds, bees live off the sun and other things without people coming togetherâ (DA_KM, Period III)
We will touch upon generic reference in the final section of this chapter, but for a more in-depth account of genericity in North Germanic the reader is directed to Skrzypek and Kurek (2018) and Skrzypek, Kurek-Przybilski and Piotrowska (2020).
5.4 Grammaticalization of the Indefinite Article
We noted in Chapter 1 (section 1.1; see also section 2.3) that the model of indefinite article grammaticalization has been tested against data from a number of languages and has been shown to predict correctly the development of the article from the numeral âoneâ (see also Brandtler and Delsing 2010, Skrzypek 2012 and 2013 for an account of the development in Swedish). With respect to the indefinite article, there is one major difference amongst the North Germanic languages, namely that the indefinite article has not evolved in Icelandic, although it is present in all other North Germanic languages, including Faroese, an insular language which has retained a number of archaic traits and is typologically similar to Icelandic.
In this section we will therefore briefly present the data from Danish and Swedish, which illustrate the development of the indefinite article in these languages. We will also consider the Icelandic data to determine whether grammaticalization of the indefinite article was initiated in that language, or whether there were no such tendencies in the studied time-frame.
5.4.1 The Presentative Marker
The first article-like function of the numeral âoneâ is that of a presentative marker, i.e., a signal to the hearer that the introduced discourse referent will be salient, and will be referred to at least once more in the remaining text. It should be noted that it is not always easy to classify a given use of the numeral âoneâ unequivocally as numeral or presentative. We chose to treat all instances of âoneâ which were not contrasted with a different numeral or quantifier as presentative. In (206) and (208), âoneâ is contrasted with âallâ or âbothâ and is thus treated as a numeral.
(206)
En
þau
váru
öll
Ã
svefni
Ã
skemmu
einni,
Ãorbjörn
ok
syn-ir
hans
ok
ÃórdÃs
and
they
were
all
in
sleep
in
bedroom
en
Thorbjörn
and
son-pl
his
and
Thordis
âAnd they were all asleep in one bedroom, Thorbjörn and his sons and Thordis.â (IS_Gis, Period III)
(207)
Og
er
þau
komu
á
Eyri
settust
þau
niður
á
einn
stól
fyrir
Ãorgeiri
og
Urðarköttur
milli
þeirra
and
as
they
came
to
Eyri
seat.refl
they
down
on
en
chair
for
Thorgeir
and
Urðarköttur
between
them
âAnd as they came to Eyri, they were seated on one chair with Thorgeir and Urðarköttur [Stone-cat] between them.â (IS_Finn, Period II)
(208)
ællær
at
the
brænnæ
bathæ
i
et
hus.
or
that
they
burn
both
in
en
house
âor that they burn both in one house.â (DA_VL, Period I)
The presentative use is therefore not so much about the numeral value, but rather about the topicality of the new discourse referent. The incipient indefinite article is typically found at the beginning of a tale, introducing its main hero or heroine, as in (209).
(209)
Thet
war
en
timæ
en
hellugh
møø
aff
tyræ
ther
hob
hafdæ
til
Vorherræ [â¦]
There
was
en
time
en
holy
maid
of
Tyr
who
hope
had
to
Our.Lord
âThere was one time a holy maid of Tyr who hoped to Our Lord [â¦].â (DA_Kerst, Period II)
However, we notice a change in the placement of the presentative NP. In the oldest extant texts (1200â1350), they were placed clause-initially, as in examples (210â212).
(210)
En
riddar
rik
ok
væl
burin
gaf-s
i
clostar
en
knight
rich
and
well
born
gave-refl
in
monastery
âA rich and well-born knight was given in a monastery.â (SV_Bur, Period I)
(211)
En
diäkne
war
j
enom
stadh
som
heet
montepessolanus
en
deacon
was
in
en
city
which
was.called
Montepessolanus
âThere was a deacon in a city called Montepessolanus.â (SV_Järt, Period II)
(212)
Kona
ein
fór
of
morgun
snemma
að
kanna
fjöru
þvÃ
að
búandi
hennar
var
eigi
heima.
woman
en
went
of
morning
early
to
search
shore
for
that
husband
her
was
not
home
âThere was a woman who went out early one morning in search of her husband who was not at home.â (IS_Jart, Period I)
In Modern Swedish and Danish clause-initial indefinite NPs are avoided, unless they are adverbials of time or place, such as en dag, which translates naturally as âone dayâ rather than âa dayâ. If the subject is indefinite, usually some other element is placed clause-initially, e.g., an adverbial, or a presentative construction is used, as in (213).
(213)
a.
?En
katt
satt
i
trappa-n.
indf
cat
sat
on
stairs-def
âA cat sat on the stairs.â
b.
I
trappa-n
satt
en
katt.
on
stairs-def
sat
indf
cat
âOn the stairs sat a cat.â
c.
Det
satt
en
katt
i
trappa-n.
there
sat
indf
cat
on
stairs-def
âThere sat a cat on the stairs.â
Interestingly enough, clause-initial indefinite NPs are found in Swedish in both Period I and II, but gradually disappear in favour of other presentative constructions, as in examples (214â216).
(214)
War
en
iuþe
ioachim
at
namne
i
nazareth
was
en
Jew
Joachim
to
name
in
Nazareth
âThere was a Jew whose name was Joachim in Nazareth.â (SV_Bur, Period I)
(215)
Thät
war
en
riddare
som
het
troillus
there
was
en
knight
who
was.called
Troillus
âThere was a knight who was called Troillus.â (SV_Järt, Period II)
(216)
I
them
daghomen
war
en
hälogher
biskopp-er
i
en
stadh
som
heth
traiectum
in
those
days
was
en
holy
bishop-nom
in
en
city
that
was.called
Traiectum
âIn those days there was a holy bishop in a city that was called Traiectum.â (SV_Linc, Period III)
We found no clause-initial presentative indefinite NPs in Danish, not even in Period I, as examples (217) and (218) illustrate.
(217)
Thet
var
en
timæ
en
værliz
man
oc
han
haffde
en
dotter
there
was
en
time
en
worldly
man
and
he
had
en
daughter
âThere was one time a worldly man and he had a daughter.â (DA_Mar, Period II)
(218)
Saa
wor
ther
een
presth
han
heet
abrathar
so
was
there
en
priest
he
was.called
Abrathar
âSo was there a priest he was called Abrathar.â (DA_Jesu, Period III)
In Icelandic the presentative use of EN is limited to Periods I and II, as shown in examples (219) and (220).
(219)
Við
þann
stein-boga
er
kastali
einn,
er
heitir
Briktan
by
this
stone-arch
is
castle
en
which
is.called
Briktan
âBy this stonearch, there is a castle which is called Briktan.â (IS_Did, Period I)
(220)
GrÃmur
hét
einn
bóndi,
mikils
hátt-ar
og
vel
járeigandi.
Grim
was.called
en
yeoman
great
kind-gen
and
well
sheep.owner
âGrim was called a yeoman, he was of great kind and a good sheep owner.â (IS_Jart, Period I)
(221)
Var
Ã
þessu
liði
ein
blóð-sjúk
kona,
sú
er
hann
segir
eftir
orðum
AmbrósÃus-ar
Mörtu
verið
hafa
was
in
this
group
en
blood-sick
woman
this
which
he
says
after
words
Ambrosius-gen
Marta
be
have
âThere was in this group a blood-sick woman, which, according to Ambrosiusâ words, was Marta.â (IS_Marta, Period II)
We find the numeral âoneâ used presentatively in all languages studied, but in Icelandic the examples are limited to Periods I and II, as in examples (219â221). Otherwise, it is common to present new referents by BNs, as in (222) below.
(222)
Maðr
hét
Ãlvir
inn
hvÃti.
Man
was.called
Ãlvir
def
white
âA man was called Ãlvir the white.â (IS_To, Period I)
While it is not impossible that âoneâ continued to be used as a presentative marker in Icelandic (the lack of examples in our corpus does not rule it out), it does not seem to have become a marker that is used with new, salient discourse referents; rather, it remained optional and never proceeded to the next stage of grammaticalization.
The Swedish data reveal that âoneâ was first used as a presentative marker under specific circumstances, i.e., clause-initially and with subjects (see also Skrzypek 2013). As early as in Period II the clause-initial position was no longer the only option, and presentative NPs appear later in sentences as well as in first position. The Danish data do not exhibit a similar phase in the development, since there are no presentative indefinite NPs placed clause-initially in the corpus. In the material, we have only found examples of presentative constructions similar to the modern ones.
5.4.2 The Specificity Marker
After grammaticalizing as the presentative marker, the incipient indefinite article may be used to introduce any new referent into the discourse, irrespective of its salience or topicality. We find occasional instances of such use of the original numeral âoneâ as early as in Period I (examples 223â224), but it is in Period II that new discourse referents are first generally presented as indefinite NPs in Danish and Swedish (examples 225â230).
(223)
Innæn
en
stath
i
thæt
land
tha
hørthæ
en
ærlik
conæ
en
strætæ
til
oc
wæl
hus-æt.
within
en
town
in
this
land
there
belonged
en
honest
woman
en
street
to
and
well
house-def
âIn a town in this land, there a street belonged to an honest woman and also the house.â (DA_ML, Period I, the only example of specificity in Period I in Danish)
(224)
oc
kombir
j
mesopotamia
landh
j
ena
borgh
som
aatte
nakor
abraham-s
brodhir
and
comes
in
Mesopotamia
land
in
en
castle
which
owned
some
Abraham-gen
brother
âAnd comes to Mesopotamia land to a castle which one of Abrahamâs brothers owned.â (SV_Pent, Period I, the only example of specificity in Period I in Swedish)
(225)
Tha
lod
vrbanus
hennis
fadær
muræ
vp
eth
torn
then
let
Urbanus
her
father
build
up
en
tower
âThen let Urbanus her father build up a tower.â (DA_Kerst, Period II)
(226)
Hoss
thæt
closter
var
haffuer
oc
eth
market
var
thære
hoos
thre
milæ
fran
closter
by
this
monastery
was
sea
and
en
market
was
there
by
three
miles
from
monastery
âBy this monastery, there was a sea and a market was there three miles from the monastery.â (DA_Mar, Period II)
(227)
Nu
j
the
samu
stund-inne
tha
gingo
twe
brödhir
aff
thy
sama
klostr-ino
j
en-om
skoogh
som
la-def
näär
thy
klostr-eno.
now
in
def
same
hour-def
then
went
two
brothers
of
def
same
monastery-def
in
en-dat
forest
which
lay
near
def
monastery-def
âNow at the same time, went two brothers from the same monastery in a forest which lay near this monastery.â (SV_Järt, Period II)
(228)
Ok
gik
ginstan
bort
til
en
biscop
ok
skriptadhe
sik
mz
storom
angir
and
went
instantly
away
to
en
bishop
and
confessed
himself
with
great
remorse
âAnd went instantly away to a bishop and confessed with great remorse.â (SV_Järt, Period II)
(229)
Tha
satto
the
hona
vth
aff
skip-it
oppa
eet
litit
skær
som
laa
mit
j
hafw-it
then
sat
they
her
out
of
ship-def
upon
en
small
islet
which
lay
middle
in
sea-def
âThen they forced her out of the ship upon a small islet which lay in the middle of the sea.â (SV_ST, Period II)
(230)
Ok
aff
henna
root
wäxte
op
en
litin
gran
aff
hwilke
borghare-n
gladdis
and
of
her
root
grew
up
en
small
spruce
of
which
citizen-def
gladdened
âAnd out of the root grew up a small spruce of which the citizen was very glad.â (SV_SVM, Period II)
In Period III, in both Danish and Swedish, all new discourse referents (which are not anchored in the preceding discourse or unique) are introduced by means of indefinite NPs, as in (231) and (232).
(231)
oc
thedhe
henne
eth
bliith
anleth
meth
ale
the
thegn
som
hwn
kunne
gøre
and
showed
her
en
mild
face
with
all
def
signs
which
she
could
do
âAnd showed her a mild face with all the signs which she could make.â (DA_Kat, Period III)
(232)
han
hauer
en
godan
skiold
then
will
jak
haffua
he
has
en
good
shield
this
want
I
have
âHe has a good shield that I want to have.â (SV_Did, Period III)
While it is not always easy to tell the presentative and specific uses of the incipient indefinite article apart, we may note that the presentative NPs are overwhelmingly animate, singular, and masculine subjects. Those NPs have the same semantic make-up as the first NPs to appear with the definite article. Here we find a parallel between the two developments. The specific use, on the other hand, is not limited in the same way, and here we also find NPs that are inanimate, plural, neuter and so forth. The discourse referents that are introduced in this way form background for the main agents in the text.
So far, we have discussed the Danish and Swedish data alone. The Icelandic data differ significantly. As we have indicated in Chapter 4 (section 4.2, Table 23), the frequency of the numeral âoneâ remains stable throughout the three periods studied, in contrast to Danish and Swedish, where we observed a noticeable rise in frequency, which often accompanies grammaticalization of an item. In the usage of the numeral âoneâ in Icelandic, we find occasional examples which could be classified as specific, i.e., introducing a new discourse referent which is neither salient nor topical. Some such examples are presented in (233â234) below.
(233)
Er
önnur
löng
leið
ok
dálig,
en
önnur
miklu
skemmri
ok
betri.
En
á
þessi
inni
skemmri
er
annmarki
nokkurr
svá.
Ãat
er
á
ein,
er
eigi
má
yfir
koma-st
nema
at
einum
stein-boga.
Við
þann
stein-boga
er
kastali
einn,
er
heitir
Briktan.
Ãann
kastala
halda
tólf
skotmenn
is
other
long
road
and
bad
and
other
much
shorter
and
better
and
on
this
def
shorter
is
obstacle
some
so
that
is
river
en
which
not
may
over
come-dep
but
by
en
stone-arch
by
that
stone-arch
is
castle
en
which
is.called
Briktan
that
castle
holds
twelve
shooters
âThere is one long and bad road, and another much shorter and better. And on the shorter one there are some obstacles. There is one river, which may not be traversed, but for a place by one stone arch. By this stone arch, there is one castle which is called Briktan. This castle holds twelve shooters.â (IS_Did, Period I)
Note that both the stone arch and the castle are presented as indefinite and are picked up later in the text; therefore, the use of en âoneâ could be taken to be presentative (the subsequent mentions are topicalized). The numeral could be pre- or postposed, seemingly with no difference in meaning (compare example 148 section 5.3.1 on direct anaphora in Polish). The NP á ein âriver oneâ introduces a new discourse referent which is not mentioned again, and the numeral could be construed as an example of a specificity marker. Similarly, in example (234) from Period II, the new discourse referent does not appear in subsequent mentions.
(234)
Kolur
sagði
þá:
Sá
eg
þá
bræður,
Helga
og
Gunnar,
Ã
hell-i
ein-um
norður
með
á-nni.
Kolur
said
then
saw
I
them
brothers
Helgi
and
Gunnar
in
cave-dat
en-dat
north
with
river-def.dat
âKolur said then: I saw these brothers, Helgi and Gunnar, in a cave north downstream.â (IS_Gun, Period II)
However, these two examples are the only ones found in the entire Icelandic corpus where the numeral âoneâ is seemingly used as a specificity marker. No other examples of such uses were found in the material chosen for this study.
5.4.3 The Non-specificity Marker
The use of EN with non-specific discourse referents is occasionally found in Period II, but it is first in Period III that we find more examples. The typical contexts include similes (comparisons, e.g., as bright as a sun), conditionals, negations, etc. Examples of the incipient indefinite article used in these contexts are quoted in (235â239).
(235)
Hwar
meghet
meræ
bær
menneske
at
arbeydhe
for
hwer
andhre-s
siele-s
salighet
at
swadanth
eth
wæænth
creaturæ
skall
ey
forkaste-s
how
much
more
must
man
to
work
for
each
other-gen
soul-gen
holiness
that
such
en
beautiful
creature
shall
not
reject-refl
âHow much more must people work for each otherâs soulâs holiness so that such a beautiful creature should not be rejected.â (DA_Kat, Period III)
(236)
færekude
swaret
hwore
motthæ
en
jomffrv
barn
føde
Farekunde
answered
how
could
en
virgin
child
bear
âFarekunde answered: How could a virgin bear a child?â (DA_KM, Period III)
(237)
ok
diäfwl-en
vppinbaradhi-s
j
enna
quinn-o
liknilse
ok
synti-s
hanum
idhkelika
j
tholiko
liknilse
and
devil-def
revealed-refl
in
en
woman-gen
likeness
and
showed-refl
him
incessant
in
such
likeness
âAnd the devil revealed himself in one womanâs form and showed himself to him incessantly in such form.â (SV_Järt, Period II)
(238)
Tha
gör
iak
som
en
folsker
man
än
frankis
land
skal
tappa
sit
loff
fore
mina
skuld
then
do
I
as
en
false
man
if
French
land
shall
lose
its.refl
praise
for
my
sake
âThen I act like a false man if France shall lose its praise for my sake.â (SV_KM, Period II)
(239)
hans
nampn
war
willebrordus,
han
aktadhe
at
visitera
eth
iomffrw
closter,
huilket
han
siälffwer
fordhom
haffdhe
bygth
oc
stiktat
his
name
was
Willebrordus
he
intended
to
visit
en
maiden
monastery
which
he
himself
before
had
built
and
established
âHis name was Willebrordus, he intended to visit a female monastery, which he himself had built and established.â (SV_Linc, Period III)
In the Icelandic data we found no examples of the use of EN in which it could be classified as a non-specific marker.
5.4.4 Summary
The model of the indefinite article grammaticalization is well-confirmed by the Danish and Swedish data. The interesting fact, however, is that the topicality of the new referent is also marked by other means, not only the incipient indefinite article, but also by the subject role and position in the sentence. Further, it is interesting to note that, despite their close relationship, Icelandic exhibits a different development than Danish and Swedish. There are isolated uses of the numeral EN in Icelandic which could be treated as presentative, but no further developments can be discerned. Indeed, it seems that some development towards an indefinite article first occurs later in the history of Icelandic history (KliÅ 2019), but it never quite takes off.
As with the definite article, we have found occasional instances of indefinite NPs used with generic reference; see examples (240) and (241). They all come from Period III and mark the final stage of indefinite article grammaticalization.
(240)
Jtem
kære
fadher
haffde
i
seeth
eller
kunne
see
een
mænneske-s
sieell
the
thwiffler
mik
enckthet
ther
paa
therefore
dear
father
had
you
seen
or
could
see
en
man-gen
soul
that
doubt
me
nothing
there
on
âTherefore, dear father, had you seen or could see a manâs soul, you would not doubt me on that.â (DA_Kat, Period III)
(241)
Enghen
thingh
er
j
wærdh-en
ther
kan
ighne-s
with
een
sieel-s
delighet
no
thing
is
in
world-def
which
can
liken-refl
with
en
soul-gen
beauty
âNothing in this world can be compared with the beauty of a soul.â (DA_Kat, Period III)
They all come from Period III and mark the final stage of indefinite article grammaticalization. This indicates that the grammaticalization of the indefinite article has by that time progressed to the final stage, however, generics may also appear as BNs at that time.
5.5 Grammaticalization of DefinitenessâA Larger Chain
At the beginning of our study, we pointed out a number of puzzles that remain unsolved in research on the diachrony of (in)definiteness. Apart from the puzzles concerning the stages of development of each article, we also find that the interplay between the two developments is far from clear.
Typological data suggest that if a language has grammaticalized the indefinite article, it is likely to have grammaticalized the definite as well. The reverse does not seem to hold, i.e., we find article languages with only the definite article, where the morphological zero is used in indefinite contexts. Diachronically, there is ample evidence that the grammaticalization of the definite article precedes the grammaticalization of the indefinite. It would thus seem that the two developments are interconnected in such a way that, when the definite article grammaticalizes, the indefinite may, but need not, grammaticalize as well.
As we have demonstrated above, the origins of both grammaticalizations lie in textual use: the demonstrative as a direct anaphoric marker, and the numeral âoneâ as a presentative marker. While direct anaphora typically points backwards, instructing the hearer to seek an antecedent in the preceding discourse, the presentative markerâs function is to arrest the attention of the hearer and direct it towards a new but salient or topical discourse referent. In this sense the presentative marker may be said to be cataphoric, as it directs the hearer forward in the text.
It should be noted that, as the incipient definite article progresses on the grammaticalization scale to mark all known discourse referents and not just topical ones, so does the incipient indefinite article come to mark all new discourse referents. In this sense the two grammaticalizations go hand in hand, and their joint result is a system of marking of specific, textually-anchored discourse referents which are accessible to the hearer through the discourse alone. Nevertheless, as the Icelandic case indicates, at this stage the grammaticalization of the indefinite article is an option, but not a necessity. We therefore have a number of languages which have grammaticalized the definite article only.
Once we leave the textual ground and come to other uses of the definite article, in larger situation use, there is no obvious parallel between the development of the definite article and that of the indefinite. However, both grammaticalizations conclude with the same article usage, i.e., in generic reference. Genericity, as the final stage of grammaticalization of each article, seems to neutralize the definiteâindefinite distinction.
At this point we must note that, despite the similarities in their developments with respect to article grammaticalization, modern North Germanic languages exhibit surprising discrepancies in article use with generic referents. Even though all NP types can be used generically (BNs, defNPs in singular and plural, plural NPs and, in the Continental languages, indefNPs), each language has its own preferences. In Norwegian (bokmål) the default form of generic NP is BN, while Danish exhibits a strong preference for singular defNPs, with Swedish results giving a multifaceted picture, though with a preference for defNPs as well (Skrzypek, Kurek-Przybilski and Piotrowska 2020).
This suggests that even though the grammaticalization models present some universal tendencies, they do not exclude language-specific developments, which may result in idiosyncratic distribution of the articles. For instance, the definite articles in some Romance languages acquire at an early stage, from the 12th century onwards, a generic interpretation in contexts where this is still impossible for English, German, and Dutch (e.g., English (*The) Whales are mammals vs. French Les baleines sont des mammifères; De Mulder and Carlier 2011:534).
If both articles evolve in a language, the definite seems to always predate the indefinite (which may not grammaticalize at all, as in Icelandic), although the two developments may overlap to some extent. In Danish and Swedish the grammaticalization of the definite article is not complete before the grammaticalization of the indefinite is initiated (see also Skrzypek 2012). As a result, the development of the definite article is at least partly conditioned by the development of the indefinite. In the absence of the indefinite article, the definite is also more likely to evolve into a specific article (Greenberg 1991, De Mulder and Carlier 2011:525), though as the Icelandic example indicates, again this is not a necessity.
Finally, both grammaticalizations involve a gradual reduction in the scope of the use of BNs. In a language with no articlesâmost Slavic languages, for exampleâBNs are the default NP forms, irrespective of intended definite or indefinite meaning. In a language with just the definite but no indefinite article, e.g., Icelandic, BNs are used where Danish, Swedish and Norwegian would use the indefinite article. However, as noted in Borthen (2003) and Asudeh and Mikkelsen (2000), the use of BNs in Mainland Scandinavian languages exceeds the use of BNs in other Germanic languages, such as English or German. BNs seem to have undergone a secondary grammaticalization (see also Berezowski 2009, Rosén and Borthen 2017, Kinn 2019), in contrast to the definite and indefinite articles, whereby the articles can be omitted to create a non-morphic, non-specific effect.
5.6 Excursus: The Incipient Indefinite Article in Icelandic beyond 1500
The data retrieved from our corpus includes examples of the presentative use of the numeral âoneâ in Icelandic and some instances of its use as a specific marker, but the development as documented here does not seem to go beyond the first stage of grammaticalization. As we have noted, the two grammaticalizations, of the definite and of the indefinite article, are interdependentâor more correctly, the grammaticalization of the indefinite article follows that of the definite, initially mirroring it in terms of the factors affecting the selection of the grammaticalizing item. We have also concluded that while the Continental languages Danish and Swedish have developed the indefinite article, the Insular language Icelandic has not. The question remains open as to whether the rise of the indefinite article is in any way aligned with that of the definite; in other words, whether there is in fact a âwindow of opportunityâ, a timeframe in which the grammaticalization of the indefinite is possible while the definite is still at the early stages of grammaticalization, or whether the indefinite article can develop at any moment in the history of a language. We have the benefit of hindsight knowing that Modern Icelandic does not possess an indefinite article; however, it would be interesting to extend our search for possible grammaticalization further than the present study allows, to establish whether there were in fact any developments toward indefinite article grammaticalization.
A claim to this effect has been made in the literature before, in particular by Leijström (1934). Previous research has also revealed an increased frequency of einn âoneâ in Old Icelandic (1200â1400). The increase reported is from 0.77 instances per 1000 words in the 13th century to 1.27 in the 15th century (Rögnvaldsson et al. 2012, Table 2). However, from the 16th to the 19th century the frequency is decidedly lower (Rögnvaldsson et al. 2012, Table 3). Although a rise in the frequency of an item is not in itself a diagnostic of the grammaticalization of that item, it often accompanies changes in its use, which may be part of the grammaticalization process. Therefore, this rise does not seem irrelevant. Our corpus, which spans the years 1200â1500, reveals similar results: very similar values in Period I (1200â1350) and Period III (1450â1550), but double those values in Period II (1350â1450) (see Chapter 4.2). As we have found in our analysis, the most likely reason for this increase is the use of einn as a presentative marker. Rögnvaldsson et al. (2012) do not analyse the uses of einn, so we have no way of comparing our data with theirs. It has been noted in the literature that the function of einn was not merely cardinal (ein, ikkje to âone, not twoâ), but also deictic: ein eller annan, ein viss âone or another, a certainâ (Haugen 2006:268). This demonstrates the formâs potential to develop into an indefinite article. However, it was not obligatory in the latter types of contexts.
Interestingly enough, there is hardly any account of einn in Icelandic from a diachronic perspective. It seems that, based on its use in Old Icelandic and the lack of an indefinite article in Modern Icelandic, the history of einn has not come under the spotlight of historical linguistics. The only major work that touches upon the subject is the now somewhat dated 1934 monograph by Gunnar Leijström Om obestämda artikeln: ett bidrag till nordisk sprÃ¥khistoria âOn the indefinite article: a contribution to Nordic language historyâ, to which we will return shortly. It is noted in the literature that einn may have the indefinite meaning if it is postposed, for example einn maður means âone manâ (cardinal meaning), while maður einn means âsome manâ, corresponding to the indefinite pronouns einhverr âsome, someoneâ and nokkur âsomeâ, for which it may be exchanged; compare with Swedish nÃ¥gon âsomeâ (Bandle 1956:330; see also Nivre 2002).
However, Leijström notes that during the Reformation, from ca. 1520, there is evidence of the use of einn in the written standard in article-like functions. He attributes this to Danish and German influence (Leijström 1934:71). The article-like use is found mainly in religious prose, and Leijström quotes relevant passages from the New Testament:
(242)
þu
ert
einn
Iude
enn
eg
em
ein
samuersk
kona
you
are
en
Jew
and
I
am
en
Samaritan
woman
âYou are a Jew and I am a Samaritan woman.â (John 4.9, Helgason 1929:116)
This use of (preposed) einn persisted until the early 1800s (Bandle et al. 2002:1268), but beyond that period the preposed einn becomes less frequent in the written standard. The reason for this seems to lie in the strong purist tendencies of the mid-1800s, whose explicit goal was to make the language free of any foreign influence, as proclaimed in among others Fjölnir, an annually published journal (1835â1844) with ambitions to revive Icelandic national consciousness and gain support for Icelandic independence (Leijström 1934:71). Part of the process was a preference for the word order noun + einn, present in Icelandic in the first extant texts with the meaning âsome xâ, which would create an antipole for the Danish order en + noun (Kristjánsson 2009:199).
It is important to observe that the only evidence of the article-like use of the numeral einn in Icelandic comes from written texts representing high literary style. It does seem as though this use was indeed a copy of the Danish and German pattern rather than a spontaneous innovation. Leijström not only rejects the idea that such use could be an archaism from Old Icelandic times, but also draws a line between the Old Icelandic uses of einn as indefinite pronoun and the modern Continental languagesâ use of the indefinite article, at the same time attributing the rise of indefinite article in Danish and Swedish to language contact with Low German, a contact which the Icelandic community, due to its isolated geographical position, escaped (Leijström 1934:171). Leijström notes also that the endemic use of einn as indefinite pronoun is most clearly visible in the Sagas of Icelanders (Ãslendinga sögur), as this genre is quite free of foreign linguistic influence, as opposed to contemporary (samtÃðarsögur) or chivalric sagas (riddarasögur), which may have been written with Continental literature as their model (Leijström 1934:85).
There are no other major works in which Icelandic einn is considered as a potential indefinite article. Some scattered information is to be found in studies of Old Norse and Icelandic syntax. Similarly to Leijström, also WestergÃ¥rd-Nielsen (1946) dismisses the possibility of the indefinite use of einn, and finds that such use in the 17th century was a result of Danish and German influence (WestergÃ¥rd-Nielsen 1946:58), while Hanssen et al. state [u]bestemt artikkel i moderne mening har det eldre norrøne mÃ¥l ikke âthe indefinite article in the modern meaning is not found in Old Norseâ (Hanssen et al. 1975:136). As an argument against the indefinite article interpretation, it is noted that the form is not obligatory, and that there is apparently free variation between BN and einn with nouns; for example, maðr er nefndr Emundr âa man is called Emundrâ is compared with a number of examples with einn (Iversen 1972:125). This view is also found in Falk and Torp (1900). On the other hand, Nygaard admits that some occurrences of einn are better described as an indefinite article (in its early stages). It is also Nygaard who describes the preposition of einn with a new, article-like meaning (Nygaard 1905).
Interestingly, it is in more recent literature that the indefinite article interpretation is found more acceptable. Bandle considers einn to be the indefinite article in Old Icelandic (Bandle 1956:330), Einar Haugen emphasizes the formâs potential to grammaticalize into the indefinite article (Haugen 2006:268), and Haugen states that the indefinite article was on the verge of grammaticalization in the 14th century, but did not come into wide use and was gradually abandoned, the only texts with significant occurrence of einn being those translated from Middle Low German (Haugen 1976:299).
In summary, it may be said that opinions on the status of einn have evolved hand in hand with theoretical developments that allow a more dynamic interpretation of language. The fact that it is known today that the potential development towards an indefinite article did not go any further, and that there is now no indefinite article in Icelandic, does not rule out the possibility that its grammaticalization was underway at some point.
One attempt to verify this hypothesis is found in a 2015 BA thesis by Phil Beier. The author chose four Sagas of Icelanders, as these are expected to be free of foreign influence and thus include the original, endemic use of einn (Beier 2015:21; see also Kristjánsson 2009:218 and Leijström 1934:85). The texts have a largely narrative character and document historical events. The sagas chosen for the study were Egils saga Skalla-GrÃmssonar (13th century), Finnboga saga ramma and Flóamanna saga (14th century) and Fljótsdæla saga (15th century). The second of these is also part of our corpus, and the timeframe of the study is identical to ours.
Beierâs quantitative results reveal that preposed einn is on the rise in the 14th century compared with the 13th century material, but its frequency is lower again in the 15th century.
Table 63
Einn in four Sagas of Icelanders, after Beier (2015)
|
Text |
einn in preposition |
einn in postposition |
Total |
|---|---|---|---|
|
Egils saga, 13th century |
47 |
34 |
81 |
|
Finnboga saga, 14th century |
37 |
10 |
47 |
|
Flóamanna saga, 14th century |
28 |
14 |
42 |
|
Fljótsdæla saga, 15th century |
32 |
27 |
59 |
|
Total |
144 |
85 |
229 |
Although not directly reported in relation to the overall length of each text, the results nevertheless indicate that the frequency of einn rises in the 14th century in the selected corpus. Beier (2015) states that none of the uses of einn in either text can be shown to be indefinite article use; he notes, however, that some examples from Fljótsdæla saga could possibly be considered quasi-article uses.
(243)
Eg
sá
járn-súlu
eina
standa
upp
Ã
hell-i-n-um
undir
ræfur
en
við
þessa
súlu
var
bundin
ein
kona.
I
saw
iron-column
en
stand
up
in
cave-dat-def-dat
under
roof
which
by
this
column
was
bound
en
woman
âI saw an iron column in the middle of the cave which bore the caveâs roof and to which a woman was tied.â (Fljótsdæla saga, Beier 2015:32)
(244)
Ãeir
sáu
út
frá
Mýnesi
hvar
einn
stakkgarður
stóð
fram
Ã
vatn-ið
Lagarfljót.
They
saw
out
from
Mýnes
where
en
fence
stood
fore
in
water-def
Lagarfljót
âOutwards Mýnes they saw a fence protruding into the waters of Lagarfljót.â (Fljótsdæla saga, Beier 2015:32)
Both the position of einn and its individualizing meaning in ein kona âa womanâ and einn stakkgarður âa fenceâ suggest that this is a potential case of an incipient indefinite article (see also Falk and Torp 1900:61).
As we mentioned before, apart from Leijströmâs monograph, which deals with the indefinite article in all Scandinavian languages and not only Icelandic, there is hardly any research into the history of einn in Icelandic. The present study concludes with texts from the 16th century, and although it confirms the formâs potential to undergo grammaticalization, it does not indicate that such grammaticalization in fact took place. In our study, which is based on material from three languages, we would like to pay more attention to the differences between the languages which may result in different article systems. We shall therefore make a closer inspection of the Icelandic einn.2
To investigate the hypothesis that the numeral âoneâ may have progressed further in its grammaticalization into an indefinite article, we have prepared a complementary corpus of Icelandic texts, which is an excerpt from The Icelandic Parsed Historical Corpus (IcePaHC; Wallenberg et al. 2011). IcePaHC includes a number of Icelandic texts, representing a variety of genres, written between 1100 and 2000. The corpus is still being updated; for the purpose of the present study, we used the version 0.9 from 2011.
The texts chosen for this supplementary study represent different periods and genres. A short description of each is given below.
-
Fyrsta málfræðiritgerðin (FM), âThe First Grammatical Treatiseâ, is a study of Icelandic phonology; it was most likely composed in the 12th century and is among the first written texts in Icelandic. It is found in Codex Wormianus, which includes three other texts, with FM being the first; this is the reason for its being named First Grammatical Treatise. It represents academic prose.
-
Miðaldaævintýri (M), âMedieval Adventureâ, written ca. 1475, is a translation from English. It is a collection of morality tales and represents religious prose (Viðarsson 2007:12â13).
-
Okur: Hvörinn það er fyrirboðið og bannað à guðsorði (Okur). âUsury. Whether it is forbidden and banned in gospelâ. This text was written 1611 by Guðmundur Einarsson. It considers religious matters, but does not have a narrative character like the morality tales in M.
-
Reisubók Jóns Ãlafssonar Indiafara (RJOI), âTravel book by Jón Ãlafsson India-travellerâ. Jón Ãlafsson was a famous Icelandic traveller, mainly known for his voyage to the Danish settlement of Tranquebar (Tharangambadi) in India. The text is an autobiography written in 1660, documenting Ãlafssonâs life and travels, with the final part written by his son. The text is available from a number of sources, and has also been published in Danish and English (Neijmann 2006:219â220).
-
Nikulás KlÃm (NK) is a satirical novel by the NorwegianâDanish author Ludvig Holberg, originally published in Latin in 1741, which gave it a worldwide audience. The Icelandic version was published in 1745 in a translation from German (Viðarsson 2007:10â11).
The supplementary corpus consists of both original Icelandic texts and translations from other languages, where some influence in the choice of einn as an equivalent of the indefinite article could be expected, both M and NK being translated from article languages with developed indefinite articles. We have included FM, even though it is dated to the period already studied in our main corpus, so as to have a viable collection of similar texts representing all periods in this supplementary study of Icelandic. Translations were included as the most likely place for foreign-influenced indefinite article-like uses of einn to appear.
Each text was excerpted in its entirety using the KWIC concordance creator, and the results were then sorted manually. Taking Heineâs (1997) model of the grammaticalization of the indefinite article (see Chapter 2) and the earlier observations on the distribution of einn (among others its propensity to appear with temporal adverbials), the excerpted instances of einn were sorted into the following categories: cardinal, potential cardinal, temporal adverbial, presentative marker, specific marker, non-specific marker. The numeralâpotential numeral dichotomy served to sort the obvious cardinal uses of einn, typically when a contrasting cardinal could be found in the context, from those that included no such contrast but were nonetheless possibly cardinal and not presentative. An overview of the results is given in Table 64.
Table 64
Einn in the supplementary Icelandic corpus (KliÅ 2019:57)
|
Texts |
Frequency |
Cardinal |
Potential |
Temporal |
Presentative |
Specificity |
Non- |
Total |
|---|---|---|---|---|---|---|---|---|
|
per 1000 |
cardinal |
adverbial |
marker |
marker |
specificity |
|||
|
words |
marker |
|||||||
|
FM (1150) |
11.8 |
18 |
3 |
â |
5 |
â |
â |
26 |
|
M (1475) |
8.9 |
2 |
8 |
17 |
91 |
â |
29 |
147 |
|
Okur (1611) |
5.9 |
2 |
6 |
4 |
8 |
â |
34 |
54 |
|
RJOI (1661) |
9.0 |
4 |
10 |
38 |
51 |
â |
65 |
168 |
|
N (1745) |
7.5 |
4 |
13 |
19 |
27 |
â |
95 |
158 |
|
Total |
â |
30 |
40 |
78 |
182 |
â |
223 |
553 |
Firstly, we note that even though the frequencies per 1000 words are not markedly different for each text, there are substantial differences with respect to each type of use. The highest frequency is found in FM, the oldest text in the sample. The instances of einn in this text comprise mainly conspicuous cardinal uses and a few instances of the presentative marker. These results largely confirm the findings of the present study and the main corpus.
Secondly, the younger texts exhibit lower frequencies of einn than FM, but the occurrences of einn there belong to entirely different categories: there are a number of examples of presentative use and, surprisingly, relatively many examples of use as a non-specific marker. In fact, this latter use turns out to be the most frequent in the supplementary corpus, quite contrary to the intuitions expressed in earlier treatments of einn. The original cardinal use of einn is equally represented in each of the younger texts, ranging between 10 and 17 instances (cardinals and potential cardinals taken together). The temporal adverbials are also relatively frequently used with einn, as the results in Table 64 indicate; there are also many instances of the presentative use of einn, with the highest number in M (from 1475). Gradually, the non-specific use takes over as the dominant use of einn in Okur, RJOI and NK. As regards the the two translations, M and NK, the latter stands out in terms of the non-specific use of einn. However, there is also a considerable time gap between the two, making M an example of quite advanced development for its time.
5.6.1 Cardinal Uses
As we have discussed the cardinal usage of einn in detail based on the main corpus, we will only quote a handful of relevant examples here from the supplementary corpus.
(245)
og
gjöri
ég
einn
staf
af
báðum
and
make
I
en
letter
of
two
âAnd I spell two letters as one.â (IS_FM, 1150)
(246)
og
lætur
þau
liggja
saman
Ã
einni
sæng,
svo
lengi
að
hans
systir
varð
með
barn
and
let
them
lie
together
in
en
bed
so
long
that
his
sister
was
with
child
âAnd they let them lie together in one bed until his sister was with child.â (IS_M, 1475)
(247)
að
hjálpa
þeim
þurfuga
um
eina
mjöltunnu
fyrir
3
vættir
fiska
Ã
einkaupi
to
help
the
needy
by
en
flour.barrel
for
three
parts
fish
in
retail
âTo help the poor by selling one barrel of flour for the equivalent of several pounds fish in retail.â (IS_Okur, 1611)
(248)
að
vega
sjálfur
Ã
metum
púðr-ið
Ã
einni
skál-inni
en
lóð-ið
Ã
annarri
to
weigh
self
in
measure
powder-def
in
en
bowl-def
but
solder-def
in
other
âTo weigh by yourself the measure of the powder in one bowl but the solder in another.â (IS_RJOI, 1661)
(249)
að
allir
innbúar
þessa
pláneta
tala
ein-u
tungumál-i
that
all
inhabitants
this
planet
speak
en-dat
tongue-dat
âThat all who inhabit this planet shall speak one language.â (IS_NK, 1745)
Note that in example (248) the numeral einn is joined with the definite form of the noun skál âbowlâ. Similar co-occurrences of the numeral âoneâ with the definite forms have also been noted in Old Swedish.
(250)
Falder
klocka
rummi
nither
i
houoth
manni
böte
sokn
fore
III
marker
arfua-nom
enom
æn
han
dör
aff
falls
bell
room
down
in
head
man
pay
parish
for
three
marks
heir-dat
en
if
he
dies
of
âIf a church bell falls down on a manâs head, the parish shall pay three marks to the heir if the man dies of it.â (YVL, Period I; quoted after Skrzypek 2012:190)
This type of construction in Icelandic, including other indefinites (suman mjoð-inn âsome honey-defâ, hvern fugl-inn âeach bird-defâ), is discussed in Perridon (1989:197), as well as in Sigurðsson (2006) and Pfaff (2017). This construction is part of a larger pattern of partitive constructions, in which a definite noun is combined with a quantifier. Sigurðsson calls these Full Concord Constructions (Sigurðsson 2006) and Pfaff calls them Little Partitives (Pfaff 2017).
5.6.2 Potential Cardinal Use
The contexts in which einn is used as a potential cardinal differ from straightforward cardinal uses, as in these contexts it is not always obvious whether âoneâ is indeed the intended reading. Such contexts usually have the property that einn is not contrasted with any quantifiers. As an example of such use, we quote a passage from NK, the most recent text in the supplementary corpus.
(251)
að
þeir
væru
hinir
vitugustu
og
sanngjörnustu
dómarar,
þeir
eð
hverki
gengist
fyrir
gjöfum
né
hótunum
að
vÃkja
eina
fingurs-breidd
frá
sannleik-n-um
that
they
were
the
wisest
and
fairest
judges
their
oath
neither
goes
for
presents
nor
threats
to
depart
en
finger-breadth
from
truth-def-dat
âThat they were the wisest and fairest judges, their oath would not be broken for presents nor threats and they would not depart from the truth by one millimetre.â (IS_NK, 1745)
The context is ambiguous, it is possible that EN here is a cardinal but an indefinite article is equally possible.
5.6.3 Temporal Adverbials and Presentative Marker
The third category includes the use of einn in temporal adverbials. Such use has already been noted in previous work; Leijström states that it was quite frequent in Old Norse (Leijström 1934:39). It is not surprising that this use is absent from FM, it being an academic rather than a narrative text. Below follows a selection of examples from the remaining four texts in the supplementary corpus.
(252)
Ã
einum
tÃma
var
einn
rÃkur
maður.
on
en
time
was
en
rich
man
âOnce upon a time there was a rich man.â (IS_M, 1475)
(253)
Gleði
þeirra
ómildu
og
óguðlegu
varir
um
eitt
augna-blik
joy
their
unkind
and
ungodly
lasts
on
en
eye-blink
âTheir unkind and ungodly joy lasts only for a blink of an eye.â (IS_Okur, 1611)
(254)
Einn
tÃma
bar
svo
við
1617
um
sumar-ið,
á
ein-um
torg-deg-i
sem
var
laugardag,
að
einn
maður [â¦]
en
time
was
so
by
1617
about
summer-def
on
en-dat
market-day-dat
which
was
Saturday
that
en
man.nom
âOne time, in the summer of 1617, it so happened on a market day that was Saturday, that one man [â¦]â (IS_RJOI, 1661)
(255)
sem
þeir
einu
sinni
heyra
which
they
en
time
hear
âwhich they one time hearâ (IS_NK, 1745)
The use of einn with temporal adverbials is widespread throughout all of the texts in the supplementary corpus.
(256)
Svo
segist
af
Ãkorn-a-n-um
að
hann
rennur
eftir
ein-um
mann-i.
Maður-inn
flýði
fast
undan
fyrir
hræðsl-u
sakir
so
says
of
squirrel-dat-def-dat
that
he
runs
after
en-dat
man-dat
man-def
fled
quickly
away
for
fear-obl
sake
âIt is told of the squirrel that he hunted one man. The man fled quickly for the sake of fear.â (IS_M, 1475)
(257)
Ãað
skrifa-st
um
einn
náttúru-stein
sem
Chalazias
er
nefndur
að
hann
sé
harð-ur
sem
demant
it
write-refl
about
en
nature-stone
which
chalazias
is
called
that
he
be.subj
hard-nom
as
diamond
âIt is said about a natural (magical) stone called chalazias (named by Pliny, most likely a type of quartz) that it is as hard as diamond.â (IS_Okur, 1611)
(258)
Ãar
kom
einn
maður,
Andrés
SandvÃk
að
nafn-i [â¦]
there
came
en
man.nom,
Andrés
SandvÃk
to
name-dat
âThere came a man called Andrés SandvÃk [â¦]â (IS_RJOI, 1661)
(259)
Ã
þess-um
stað
var
eitt
akademÃum
eður
gymnasium.
Ãar
voru
kenndar
frÃ-kúnstir
með
bestu
siðsemi.
in
this-dat
town
was
en
academy
or
gymnasium
there
were
known
free-arts
with
best
propriety
âIn this town, there was an academy or a gymnasium. Free arts with highest propriety were famous there.â (IS_NK, 1745)
There is, however, no reason to suppose that the use of EN here is a diagnostic of progressing indefinite article grammaticalization. Nor does the abundant presentative use seem to be such a diagnostic, as it is also found throughout the texts studied. Some examples of the presentative marker follow below. Note that no such examples were found in FM.
5.6.4 Marker of Specificity
The marker of specific reference cannot be easily distinguished from the presentative marker, as both introduce new discourse referents. The only difference seems to lie in the discourse referentâs topicality or lack thereof. Discourse referents introduced by the presentative marker are salient and likely to be anaphorically referred to later in the text, while those introduced by the specificity marker may be backgrounded and need not receive any subsequent mentions. By this definition, if the discourse referent is mentioned in subsequent text, the marker used to introduce it (if any) may be classified as presentative; if not, it is classed as specific. This distinction is extremely crude; taking the presence of subsequent mentions as the only feature distinguishing the two uses, we located only two examples of the specific use of einn in the supplementary corpus.
(260)
Ã
ein-um
stað
er
Lundún
heitir
Ã
Englandi
bar
svo
til
að
einn
rÃk-ur
maður
og
annar
órÃk-ur
kærðust
við
um
eitt
lÃtið
land
in
en-dat
town
which
London
is.named
in
England
carried
so
to
that
en
rich-nom
man.nom
and
other
unrich-nom
quarrelled
with
about
en
small
land
âIn a town called London which lies in England it so happened that a rich and a poor man quarrelled over a small piece of land.â (IS_M, 1475)
Neither discourse referent is referred to in the remaining text, based on this we could argue the use of EN to be as marker of specificity rather than a presentative one.
5.6.5 Marker of Non-specificity
As we have discussed in Chapter 2, the spread of the numeral âoneâ to non-specific contexts is a turning point in the grammaticalization of the indefinite article, since the numeral cannot usually be used in such a context. What we refer to as a ânon-specific contextâ includes in fact a number of different contexts. Similarly to indirect anaphora, with such heterogeneity we expect some of the contexts to admit the incipient indefinite article sooner than others.
One such context is similes. It has been noted previously that einn can be found in similes relatively often (Leijström 1934; see also Skafte Jensen 2016 for similar results in Danish), and this observation is borne out by the data from the supplementary corpus. While no such examples could be found in FM, the other texts include a number of relevant instances.
(261)
Hann
kallaði
til
sÃn
einn
fjanda,
hver
eð
vera
skyldi
svo
sem
hans
meistari,
hafandi
á
sÃnu
höfði
eina
kórónu,
og
eitt
tré
Ã
hendi-n-ni,
og
sitja
á
stól-i
sem
einn
kóngur.
he
called
to
self
en
devil
which
oath
be
guilty
so
as
his
master
having
on
his
head
en
crown
and
en
tree
in
hand-def-dat
and
sit
on
chair-dat
as
en
king
âHe summoned a devil who owed him an oath as to his master, who had a crown on his head, a tree in his hand and sat on a chair as a king.â (IS_M, 1475)
(262)
ÃvÃ
þar
evangelium
predika-st,
þar
er
mesta
uggleysi
hjá
þeim
sem
hlýða
eiga,
ein-s
og
Ã
Nóa
tÃð
og
Loth-s,
og
allmörgum
af
þeim
er
svo
þvers
um
geð
kristilega
að
lýsa,
sem
ein-um
soltn-um
hund-i
að
fasta.
therefore
where
Evangelia
preach-refl
there
is
most
fearless
by
them
who
obey
must
en-gen
and
in
Noah
time
and
Lot-gen
and
some
of
them
are
so
oppose
about
temperament
Christian
to
proclaim
as
en-dat
hungry-dat
dog-dat
to
fast
âThus, where the gospel is preached, there is the least fear in those who obey and in the time of Noah and Lot, and some of them are as much against claiming the Christian temperament as a hungry dog is opposed to fasting.â (IS_Okur, 1611)
(263)
hversu
þær
gæti
gjört
mig
sem
lÃkastan
einni
eik
how
they
could
made
me
as
like
en
oak
âHow they could make me like an oak.â (IS_NK, 1745)
(264)
þá
sýndist
þú
öll-um
oss
sem
ein
ónýt
jarðar-byrð-i
then
seemed
you
all-dat
us
as
en
unnecessary
earth-burden-dat
âThen you seemed to all of us as an unnecessary earthly burden.â (IS_NK, 1745)
Admittedly, the use of einn in this particular context is perhaps in itself no diagnostic of indefinite article grammaticalization either, as we occasionally find indefinite pronouns here. In this sense, one could argue that the use of einn in the examples above is no more than that of the indefinite pronoun. A counterargument is found in the position of einn, namely its preposition, which, according to Nygaard, is a hallmark of the beginning of grammaticalization in this article-like use of the numeral. However, we must note that similes constitute the majority of examples classified as non-specific here.
Unequivocally, the use of einn in negative polarity contexts, for example in the scope of negation, must be considered non-specific. A handful of such examples were retrieved from the supplementary corpus.
(265)
þó
þeir
geti
ei
lagt
af
við
sig
nema
eina
túbaks-pÃpu
sér
til
heilsubótar.
though
they
could
not
laid
off
by
self
but.neg
en
tobacco-pipe
self.dat
to
healing
âThough they could not quit but one pipe for the sake of health.â (IS_Okur, 1611)
As with the specific uses, there are too few instances of EN used in non-specific contexts to make any claims about the grammaticalization process. They could be isolated instances, dismissed as errors or the result of foreign language influence.
5.6.6 Other Types of Non-specific Contexts
There are a number of other non-specific contexts, such as constructions with conditionals and verbs of volition, or reported speech where the distance from the reported dialogue is marked by the subjunctive mood.
(266)
og
þess
vegna
út-send-u
eina
kvinnu
er
þeirra
lÃn-klæði
hafði
að
vakta,
og
báð-u
hana
trúlega
einn
örugga-n
og
fróma-n
man-n
að
útvega
and
this.gen
reason
out-send-pl
en
woman
which
their
linen-clothes
had
to
watch
and
asked-pl
her
presumably
en
trustworthy-acc
and
pious-acc
man-acc
to
choose
âAnd for this reason, they sent a woman who had watched their linens, and asked her to choose a trustworthy and pious man.â (IS_RJOI, 1661)
Example (266) is an opaque context, i.e., it could be interpreted as either specific (there was a certain pious and trustworthy man to be chosen) or non-specific (any such man would be eligible). The context suggests a non-specific reading.
(267)
þvÃ
þeir
meina,
að
ein
respublica
eður
landsstjórn
kunni
ómögulega
að
standast,
þar
sem
hver
og
einn
má
umbreyta
lög-u-n-um
og
aftaka
þau,
eftir
eigin
velþóknan
for
they
mean
that
en
republic
or
government
could
impossibly
to
stand
there
which
each
and
one
may
change
law-dat.pl-def-dat
and
refuse
them
after
own
gratification
âFor they mean that a republic or a government could not possibly exist where everybody could change the laws or refuse to adhere to them for their own gratification.â (IS_NK, 1745)
In example (267) a cardinal reading is possible, if we assume that the intended meaning was that a single lone republic would not survive if men were allowed to change the laws according to their whims; but it seems a far-fetched one, since it would imply a contrast between such a system and a hypothetical one with more than one republic or government, which could better weather such adverse circumstances. The most neutral and natural reading is non-specific, especially considering the presence of the negated adverb ómögulega âimpossiblyâ. A further curiosity is the presence of einn before respublica ârepublicâ (a loan word) and its absence before landstjórn âgovernmentâ. This may reflect the influence of the original German text from which the Icelandic version was translated, with a single use of the indefinite article to render the whole NP indefinite.
Specific
(268)
Ãg
minnist
hér
um
leið
graf-skrift-ar
einn-ar
yfir
bond-a
nokkur-n,
sem
þeir
Ã
Keba
gjörðu
yfir
hann,
og
svo
hljóðar [â¦]
I
remembered
here
about
way
grave-script-gen
en-gen
over
yeoman-obl
some-dat
which
they
in
Keba
made
over
him
and
which
proclaimed
âHere I was reminded of an inscription on a grave of some yeoman, which they made for him in Kebi and which said [â¦]â (IS_NK, 1745)
Observe the contrast between einn âoneâ and nokkur âsomeâ in (268) above. The specific reference is rendered by means of einnâthe speaker recalls a certain grave inscription once seen. On the other hand, the person that was commemorated by this inscription is presented as âsome yeomanâ by means of nokkur. This use of the indefinite pronoun is reminiscent of the spesumptive use of nÃ¥gon (a cognate of nokkur) in Modern Swedish, in positive polarity contexts, e.g., Jag talade med nÃ¥gon medicinsk expert. âI spoke with some medical expertâ (Nivre 2002:12, the paper also includes a discussion of these uses of nÃ¥gon and its cognates in Continental Scandinavian languages). The use of nÃ¥gon rather than the indefinite article indicates that the speaker is unable or unwilling to give further information about the referent; it also prevents the referent from being established in the discourse and referred to in the following text. The Icelandic data does not allow us to pursue the subject, but we note that two forms are chosen in the text, treating two discourse referents differentlyâthe one introduced with einn is referred to in the rest of the text, the one introduced with nokkur is not.
(269)
Soddan
kauphöndlan
við
hinn
fátæka,
nauðstadda,
kalla-st
okur
og
er
eitt
fordæðuverk
meðal
kristinna
mann-a
og
viðurstyggð
fyrir
guð-i
such
negotation
with
the
poor
necessitous
call-refl
okur
and
is
en
witchcraft
among
Christian
man-gen.pl
and
vileness
for
god-dat
âSuch negotiations with the poor, the necessitous, are called okur âusuryâ and it is a witchcraft when driven among Christian men and a vileness before God.â (IS_Okur, 1611)
Conditional
(270)
hefðu
þeir
meint,
að
ég
hefði
viljað
nauðga
einni
frú
af
þvÃ
ypparlegasta
standi
had
they
meant
that
I
had
wanted
compel
en
woman
of
this
highest
station
âThey thought that I had wanted to force a woman of the highest class.â (IS_NK, 1745)
(271)
þvÃ
mér
sýndust
þeir
heldur
spila
eina
comædiu,
en
þeir
héldi
réttargang
therefore
me
seemed
they
rather
play
en
comedy
than
they
hold
trial
âIt therefore seemed to me that they would rather play a comedy than hold a trial.â (IS_NK, 1745)
Non-referential
Finally, there are a number of examples where einn is used in a predicative NP, as (under some restrictions) the indefinite article is used in the modern Continental languages.
(272)
Severin-s
talsmann-s
dóttir
sem
er
ein
vel-talandi
jungfrú
Severin-gen
speaker-gen
daughter
who
is
en
well-spoken
woman
âThe daughter of speaker Severin, who is a well-spoken young woman.â (IS_NK, 1745)
We may observe that it is not inconceivable that the form is still a cardinal here, as one can say in English, emphatically, âshe was one fine womanâ. Similar examples from the same text are presented in (273) and (274).
(273)
ÃvÃ
hann
var
einn
tilvalinn
höfuð-dispútant
it.dat
he
was
en
appointed
head-disputant
âFor he was an appointed main disputant.â (IS_NK, 1745)
(274)
þá
getur
hann
allvel
dugað
til
að
vera
einn
hof-hlaupari
then
can
he
nevertheless
suffice
to
to
be
en
court-messenger
âNevertheless, he can then suffice to be a court messenger.â (IS_NK, 1745)
One can note, however, that such an abundant use of einn in this position points towards a more neutral rather than emphatic use. It most certainly seems also to be a case of influence of the original.
5.6.7 Generic Reference
The models of grammaticalization of both the definite and the indefinite article identify the use of the articles with generic reference as the ultimate stage of the development: in Heineâs (1997) terms, the conventionalized article. We have observed before that this final stage seems to neutralize the definiteâindefinite distinction. We have also found a number of generic uses of the numeral âoneâ in the Continental languages. However, the supplementary Icelandic corpus did not supply any such examples.
Undoubtedly, we have only touched upon the story of einn in Icelandic, given the limitations of our main and supplementary corpora. Nevertheless, we have demonstrated that the development of einn into the indefinite article, though obviously unsuccessful, was underway, at least to some extent, in the course of Icelandic history. It might have originated, as it did in the other North Germanic languages, due to internal processes, as a reaction to the early stages of the grammaticalization of the definite article, or it could have been instigated by intensive contact with languages with well-formed indefinite articles, such as Danish and German. If the former is true, then there are a number of possible explanations for why it never really took hold. Firstly, it seems that while einn undoubtedly can serve as a presentative marker, it is not obligatory in this function in texts from 1200â1550, while the definite article continues to grammaticalize and spread to new contexts. It is not inconceivable that there is a âwindow of opportunityâ for the grammaticalization of the indefinite article, defined by the stage of definite article grammaticalization. We may further consider the position of einn with respect to the main noun. Characteristically, Icelandic is not only the sole representative of the North Germanic (and all Germanic) languages which did not grammaticalize the indefinite article, it is also the only one with a very limited use of the preposed definite article. Even though it is possible to prepose the definite article hinn, it is by no means stylistically neutral. The preposed definite article can be used only if the noun is modified by an adjective (as in the other North Germanic languages); it is traditionally viewed as characteristic of formal or written Icelandic, although stylistic considerations are not the only ones; and the two forms of the article (pre- and postposed) are not completely equivalent from a semantic point of view (Thráinsson 2007:4â6). When a non-restrictive reading of a definite noun phrase is intended, the preposed definite article is required and the postposed one excluded (Sigurðsson 2006, Thráinsson 2007, Pfaff 2015, Ingason 2016b, Harðarson 2017).
If, on the other hand, the origins of indefinite article grammaticalization in Icelandic lie in language contact with Danish and German, we may easily understand how the strong purist tendencies of the 18th century might have put a stop to this development. A question that remains open in this scenario is to what extent language policies may affect language change, but with a highly literate population such as that of Iceland, it is not inconceivable that such policies may be successful.
The two scenarios are not mutually exclusive. It is possible that the potential of einn to grammaticalize was to some extent realized in the early history of Icelandic, and that at a later stage, in contact with closely related and typologically close languages (German in particular), this potential was re-activated. Whether language policy alone could be responsible for checking this development, or whether it was still at odds with the general make-up of the language, is a topic for a further, more in-depth study, taking into account spoken as well as written varieties of Icelandic.
5.7 Summary
In this chapter, we have considered the corpus material qualitatively, with particular reference to the models of article grammaticalization proposed in the literature, for definite articles arising out of demonstratives and for indefinite articles arising out of the numeral âoneâ. While the indefinite article grammaticalization model has been successfully tested against data from a number of languages, the definite article grammaticalization model still presents some puzzles, in particular that concerning the major switch from familiarity-based, strong definite semantics uses at the early stages of grammaticalization to the uniqueness-based, weak definite semantics uses at the subsequent stages of the process.
Our data suggest that the incipient definite article is first used in direct anaphoric contexts, as marker of high accessibility of the discourse referent. However, it is not impossible to use BNs anaphorically throughout Period I in all languages. It is first in Period II that their use becomes limited to referents that are unique and do not need an antecedent to be identified. We have further found that the distance between the anaphor and the antecedent, measured by a number of intervening syntagms, may affect the choice of the anaphoric marking.
We have also demonstrated that the bridging context, i.e., one that is ambiguous between direct and indirect anaphoric reading, can be found in anaphoric chains, in which the same discourse referent may find a different representation (pronominal, nominal) and the hearer may be informed of its new properties. In this sense, the reference builds both on the direct anaphora (repetition of the antecedent) and on the indirect anaphora (anchoring the anaphor in other information).
The indirect anaphora proved to be a complex and heterogenous context and the definite marking was not acquired simultaneously in all its sub-types. In particular, the data provide us with strong arguments that the inalienable possession (body parts, items of clothing), which is an example of a partâwhole relation, was marked with reflexive possessive pronouns rather than the incipient definite article as late as in Period II. Only in Period III did the use of the definite article stabilize in this type of context.
The grammaticalization of the indefinite article, as documented in the North Germanic sources, conforms to the proposed model and proceeds through the presentative use of the nominal âoneâ (to introduce new and salient discourse referents), the marking of specificity and, finally, the marking of non-specificity. This process does not take place in Icelandic. However, our study of a supplementary corpus of Icelandic texts from 1200â1745 has revealed that the numeral âoneâ can, in fact, be found in article-like uses. It can also be speculated that the grammaticalization process started but never reached the stage of the obligatory use of âoneâ in these contexts.
In the final chapter we will bring together the quantitative and the qualitative results and propose a refined model of the grammaticalization of the definite article, while discussing the interdependencies between the two grammaticalization processes.