Differences in the use of deictic expressions in
English and German texts
Article to appear in Linguistics (early 2011)
Sonderforschungsbereich 538: Mehrsprachigkeit
The article presents a contrastive analysis of the use of English and German deictic
expressions. Its focus is on the communicative role of these items, i.e. the way in which
they are used by authors to communicate effectively with their readers. The analysis tries
to combine a qualitative (discourse analytic) and a quantitative (corpus linguistic)
perspective by making use of a small corpus containing the endings of 32 English and 32
German texts from the genre popular science. All deictic expressions present in the
corpus were manually identified, counted and analyzed according to the function(s) they
fulfill in their respective context. The results suggest that deictic expressions are more
frequent in German than in English texts. Two (related) reasons seem to account for this
finding: first, deictics figure more prominently in the German system of textual cohesion.
Second, they were in many instances found to serve as an (optional) instrument for
maximizing explicitness, a communicative strategy which is customary in German but
not in English discourse.
The present article compares the use of deictic expressions in English and German
texts from a pragmatic perspective. It aims to address the different ways in which
deictic expressions are employed by English and German authors in order to
convey their ideas, interact with their readers and create textual cohesion. It will be
shown that there are considerable differences in the English and German uses of
deictic expressions, which are due not only to differences between the two language
systems, but also to the (tacit) norms and conventions which govern their use.
The current state of research does not permit a satisfactory definition of the concept
of deixis. Anderson and Keenan (1985) adopt the “standard usage” in considering
as deictics “those linguistic elements whose interpretation in simple sentences
makes essential reference to properties of the extralinguistic context of the
utterance in which they occur.” (p. 259) When we go beyond the “simple sentence”,
however, we find that this common sense definition no longer holds, because in
actual written texts, most deictics do not refer to the extralinguistic context (which
for the most part is not shared between author and reader and thus cannot be
referred to), but to the context built up by the text itself (Ehlich 2007). How about
“reference to properties of the linguistic and extralinguistic context”, then? The
problem with such a definition – as well as with similar definitions which try to
cover deictic reference in text – is that it blurs the boundary between deixis and a
related phenomenon, anaphora. The following section, which reviews some
important steps towards a theory of deixis, will thus be concerned with this
The article is structured as follows. In the next section, we will discuss Konrad
Ehlich’s approach towards a theory of deixis, which pays close attention to the
different nature of deixis and anaphora and may thus serve as a useful guide for
investigating the use of deictic expressions in text. After Sections 1.2 and 1.3 have
introduced the data and hypothesis of the present study, the results will be presented
and discussed in Section 2. The final section (Section 3), contains a short summary
of the findings as well as some general conclusions.
1.1. Towards a theory of deixis: distinguishing deixis from anaphora
This section presents Konrad Ehlich’s (approach towards a) theory of deixis. The
theory is rooted within the research tradition of Functional Pragmatics (see Redder
2008 and Rehbein and Kameyama 2006 for recent overviews). This research
tradition views language as a “sociohistorically developed action form” (Redder
2008: 136) that a speaker may use to effect changes in the hearers’s knowledge.
Ehlich’s theory of deixis was chosen as a basis for the study presented in this article
rather than another approach to coreference or deixis (see below for references)
because it pays special attention to the difference between deictic and anaphoric
expressions in terms of the cognitive processes they trigger. It will be argued that
this difference is crucial for the proper description of the use of deictics in written
discourse. Since the focus of this paper lies on the contrastive study that it presents,
this introductory section can do no more than outline the theoretical approach and
present the terminology1 to be used, thus largely avoiding comparisons with other
approaches to coreference (e.g. Ariel 1990, 2001, Gundel et al. 1993, Grosz et al.
1995, Arnold 2008) and deixis (see e.g. Herbermann 1988, Blühdorn 1995, Diewald
1991, Levinson 2004 and Sidnell 2005 for critical overviews of research on deixis).
The traditional paradigm of personal pronouns is a ragbag of two fundamentally
different groups of expressions (cf. e.g. Lyons 1977: 638–639 and the extensive
discussion in Ehlich 1979). The so-called first and second person pronouns are
prototypical members of the “Zeigfeld der Sprache” (‘deictic field of language’,
Bühler 1934), as they refer to the speaker (I, we2) and to the addressee (you) of the
respective speech situation, i.e. to entities of extralinguistic reality. They may be
termed deictic expressions, or deictics (< Gr. deiknynai ‘to show, point out’).
The so-called third person pronouns (he/she/it, they), on the other hand, do not
directly refer to the real world. Rather, they presuppose the existence of an
antecedent, i.e. an expression in the preceding discourse3 with the same referent,
with which they can thus be said to be coreferential (Halliday and Hasan 1976: 3–
4). Accordingly, these expressions may be adequately termed anaphoric
expressions, or anaphorics (< Gr. anapherein ‘to carry back’)4. In (1), for example,
all three occurrences of it (in the last case taking the possesive form its) are
coreferent with the antecedent noun phrase society.
Society reaps what it sows in the way it nurtures its children.5
As the example shows, we can say that anaphoric expressions like it ‘indirectly’
refer to the world, i.e. through their antecedent (cf. Lyons 1977: 660). It thus turns
out that the term pronoun is only suitable to describe anaphorics, which (in most
cases6) actually stand for, i.e. instead of (‘pro’), a noun, or rather a noun phrase (cf.
Lyons 1977: 636–637 on the infelicity of the term pronoun). Deictics, on the other
hand, are attention-managing devices (Ehlich 1992; a similar point is made by
Diessel 2006) that may directly refer to extralinguistic entities. They are the
linguistic correlates of attention-directing gestures such as the pointing of a finger
(“You!”)7. Whether through language or through gesture, the directing of the
addressee’s attention takes place in a demonstration space which has its origo, i.e.
its center, in the ‘here’, ‘I’ and ‘now’ of the respective speech situation (Bühler
1934). Following Ehlich (1982), we may say that while the function of deictics is to
establish a new attention focus in the addressee (by directing their attention to an
element of extralinguistic reality), the function of anaphorics is to maintain, i.e. to
carry on, an existing focus8. That is, deictic and anaphoric expressions trigger
different cognitive processes, which we will call deixis and anaphora.
So far, the difference between deictics and anaphorics is clear-cut. The confusion
begins when we take the so-called demonstrative pronouns (this/these, that/those)
into account, which we will call object deictics (following Diewald 1991: 228–
230). These may not only directly refer to the extralinguistic world (“Look at
that!”); it appears that they, like anaphorics, can also refer ‘indirectly’, as in the
To prevent the cataplectic attacks of narcolepsy, physicians can prescribe
agents that increase the availability of norepinephrine in the brain. These
include monoamine oxidase inhibitors [...].
Applied to insects, transgenic technology can offer biologists new ways to
investigate, control and exploit these creatures [...].
In (2) and (3), the deictic these is obviously coreferent with preceding linguistic
expressions, its antecedents being agents that increase [...] in (2) and insects in (3).
The only difference between the two uses of these is that in (3), its referent is pre-
categorised as creatures. In principle, the deictic could be replaced by an anaphoric
in both cases (they and them, respectively). Occurences of this/these like those
above are usually called ‘anaphoric uses of deictics’. Such terminology, however, is
problematic, as it implies that the deictic in this case triggers a focus-maintaining
(i.e. anaphoric) instead of a focus-establishing/focus-shifting (i.e. deictic) cognitive
process in the addressee. If this were true, deictics could always be replaced by
anaphorics. But as the following example shows, that is not the case.
The outflows from the sun and its stellar contemporaries blew away the
leftover gas and dust that threaded the space between them. This weakened
the gravitational glue that bound them together [...].
In (4), no antecedent expression of this can be identified. The deictic seems to have
the whole preceding sentence as its antecedent. It thus cannot be said to maintain an
existing attention focus – the meaning of the whole sentence would have to be
focused. The deictic rather establishes a new attention focus by shifting the
addressee’s attention to the state of affairs expressed in the preceding sentence (cf.
Consten et al. 2007, Consten and Knees 2008). This is why the substitution of an
anaphoric (it) for the deictic would have a confusing effect on the reader, to say the
least. It is thus not surprising that anaphoric reference to ‘higher-order entities’9
such as states of affairs or propositions is rare (Webber 1991, Hegarty et al. 2001),
while deictic reference to these entities is so common that it can be said to have a
“central function” in creating textual cohesion (Consten et al. 2007: 83). We should
also note that once a higher-order entity has been referred to by means of a deictic
like this, further reference to this entity may be made with an anaphoric (2007: 95):
cf. It also weakened... as a possible continuation of (4). This observation too is
predicted by Ehlich’s theory: once the addressee’s focus has been shifted to a
referent, the referent becomes accessible to (focus-maintaining) anaphorics.
The previous example has shown that, due to their focus-shifting nature, deictics
work better in coreferring with complex antecedents than anaphorics. The following
(constructed) example shows that in coreference with concrete antecedents the two
types of referential expressions also show a clear division of labor:
[Modern computers]i can perform [different tasks]k. Theyi / #k / These #i / k ...
(a) # Theyk include mathematical calculations, text processing...
(b) Thesek include mathematical calculations, text processing...
(c) Theyi can solve mathematical equations, process textual data...
(d) # Thesei can solve mathematical equations, process textual data...
The first sentence in (5) introduces two referents, ‘modern computers’ and
‘different tasks’. When we continue the sentence with a coreferential expression,
the anaphoric they tends to be coreferent with the subject of the sentence (modern
computers), while the deictic these tends to corefer with its object (different tasks)
(cf. the provided subscripts). Sentences (5a) through (5d), which present possible
continuations of (5), illustrate this. In these sentences the meaning of the predicate
(include... vs. solve...) forces the coreferential expression to corefer with either the
subject or the object of the preceding sentence. It turns out that when the ‘wrong’
coreferential expression is chosen, the utterance is pragmatically awkward (marked
The discussion of example (5) suggests that deictics and anaphorics are in a quasi-
complementary distribution (cf. Byron et al. 2008 for English, Bosch and Umbach
forthcoming for German), ‘quasi’ meaning that there are cases where both kinds of
coreferring expressions may be used in a semantically and pragmatically
appropriate sentence (cf. the discussion of (2) and (3) above). This conclusion is
supported by the findings of Bosch et al. (2003). In their study of a corpus tagged
for coreference relations, they found that deictics more frequently refer to non-
nominative constituents than anaphorics, which, in turn, more frequently occur with
nominative antecedents. They interpret these results as evidence for the Subject
Hypothesis, which postulates that personal pronouns prefer subject antecedents
while demonstratives prefer non-subject antecedents10. This is exactly what we
observed in example (5) and what one should expect from the difference between
anaphora and deixis as it has been outlined above, since it can be plausibly assumed
“that entities introduced in syntactically prominent positions are more likely to be
brought into focus of attention than ones introduced in a less prominent position.”
(Gundel et al. 2003: 297)11
I have argued that it is misleading to describe the function of coreferential deictics
in written discourse as ‘anaphoric’, since deictics are associated with a distinct
cognitive process, namely the shifting of the addressee’s attention focus. It is
plausible to assume that the focus-shifting function of coreferring deictics has
developed from their function of directing the addressee’s attention to objects of the
(extralinguistic) speech situation (cf. Lyons 1977: 670, Diewald 1991: 110–111).
But how exactly are the coreferential use and the extralinguistic use of deictic
expressions related? To answer this question, let us draw a more accurate picture of
deictic coreference in discourse that makes explicit its connection to extralinguistic
During discourse, be it spoken or written, the knowledge of the addressee is
updated by each utterance. This enables the author of a written text to reconstruct
the “running knowledge” of the addressee, which may then serve as a target of
deictic expressions (Blühdorn 1993, 1995), i.e. as a demonstration space in its own
right: text space (Ehlich 1983: 89; cf. Lyons’ 1977: 670 notion of a “universe-of-
discourse”). From this perspective, coreferential deixis is not qualitatively different
from deixis in (real) space or time, only the demonstration space is different (Ehlich
2007: 41). What both coreferential deictics and ‘real’ deictics have in common is
that they instruct the addressee to focus their attention on a referent already
represented in or yet to be introduced into their knowledge (cf. Ehlich 1982 and
Blühdorn 1993: 45). As for the cognitive process underlying deictic coreference, in
many cases we should expect a refocusing rather than a focusing, assuming that an
initial focusing of the addressee has already taken place (e.g. when the item referred
to was introduced into the discourse).
In this section, I have stressed the differences between deictic and anaphoric
expressions. Still, we should bear in mind that authors have considerable leeway in
deciding whether it is necessary or desirable to employ a deictic rather than e.g. an
anaphoric. This becomes particularly evident when we recall examples such as (2)
and (3), where different types of referring expressions would do the job. In general,
the choice of referring expressions in written discourse has been shown to be
crucially dependent on the communicative intention of the writer (Moya Guijarro
2006) and on the assumptions s/he makes about the knowledge of the addressee
(Arnold 2008). From this perspective deictic expressions may be seen as one
particular way of achieving addressee orientation (cf. Böttger and Probst 2001).
Now a host of contrastive studies12 has shown that English and German texts rely
on quite different strategies of addressee orientation. English and German authors
make quite different “global assumptions” about the addressee, which are an
important factor influencing the production of referring expressions (Arnold 2008).
So it makes sense to ask which role(s) deictic expressions qua attention-managing
devices play in the different strategies of addressee orientation operative in the
English and German language communities. This is the purpose of the study to be
presented in the following.
1.2. Data and scope of study
Differences in the use of deictic expressions in English and German are rarely
mentioned (but see Canavan 1972 and von Stutterheim 1997). The aim of the
present study is to obtain a first impression of possible differences in the use of
deictics in English and German texts. To achieve this, a mixed quantitative-
qualitative approach was chosen: 32 random German texts and 32 random English
texts were singled out for analysis. To ensure maximum comparability, all texts are
of the same genre (popular science) and from the same time period (1978–2002).
As it would not be feasible to manually analyze all occurrences of deictics in all 64
texts, only the last 10 orthographical sentences of each text were extracted to form a
‘mini-corpus’ of text endings. The structure of the corpus is outlined in Table 1.
Table 1. Structure of the corpus of text endings
32 text endings
32 text endings
It is no coincidence that the endings of the texts were chosen as a basis for the
corpus to be compiled rather than their beginnings or middle parts. First of all, it
was found that the text beginnings contain only few object deictics, since the
introductory parts of popular science texts seem to be mainly concerned with
introducing new referents rather than with referring to previously introduced ones.
The middle parts of the texts, on the other hand, were judged to be quite
heterogenous (and therefore unsuitable for the compilation of a small-scale corpus),
since they feature a variety of different discourse types such as short narratives or
direct speech. In contrast, the text endings were found to be quite homogenous;
they generally feature a short outlook that highlights the relevance of the topic of
the article to the reader or to society as a whole (cf. section 2.1.1 and the examples
The reader is reminded, however, that text endings are not necessarily
representative of the genre as a whole – not to mention the many other genres of
English and German. Research has shown that the use of referring expressions in
general (Fox 1987, Moya Guijarro 2006) and of deictics in particular (Diewald
1991) is dependent on genre (but see Toole 1996, who argues that the same
cognitive mechanisms are operative regardless of the genre at hand). Further study
is needed to assess how far the use of deictic expressions in the investigated texts is
influenced by genre-internal conventions (as opposed to crosslinguistically different
communicative conventions). One possible way of doing this would be to repeat the
study presented here using texts from other genres.
Methodologically, the small-scale corpus used for the present study combines the
best of two worlds, as it enables a qualitative analysis of all occurring deictic
expressions in context while still being representative enough to allow for a
quantitative perspective on the data. However, statistical significance was not
computed in order to highlight the tentative character of the quantitative findings. It
is not the prime goal of this study to deliver reliable quantitative results (the
investigated corpus would be too small for that anyway) but rather to provide
directions for further research by identifying items which seem worthwhile for
detailed contrastive investigations on the basis of larger corpora (e.g. English then
vs. German dann, cf. Section 2.4).
Ehlich suggests that the high degree of complexity in German scientific texts (cf.
Fabricius-Hansen 2000) “implies the use of a large variety of deictic expressions
being used for text organizing deictic procedures” (Ehlich 1992: 224) and that
“English seems to have chosen different strategies for representing complexity and
abstractness in scientific texts. So English seems to be poor in deixis, German rich,
and French even richer.” (p. 225) Ehlich’s suggestions led to the formulation of the
following two hypotheses for the present study:
1. The investigated German text endings make more frequent use of deictic
expressions than their English counterparts.
2. The reason for the frequency difference is a stronger reliance in German on
deictics for the establishment of textual cohesion.
The hypotheses build upon each other: while the first hypothesis postulates a
certain quantitative result, the second one makes a claim about its qualitative
The deictic expressions13 identified in the corpus are listed in Table 2. The
classification underlying the table is deliberately heterogenous: quality deixis is
really a subclass of object deixis (see Section 2.3); and composite deictics might as
well be distributed among the other classes. However, this would conceal an
interesting finding: while composite deictics are an important resource for German
popular science authors, their use is almost non-existent in the English text endings
(64 occurrences in the German text endings vs. a single occurrence in the English
ones). In contrast, English popular science authors seem to put much more
emphasis on personal deixis than their German colleagues. Departing from these
preliminary insights, the categories of deictic expressions listed in Table 2 will be
discussed in the following sections14.
Table 2. Types of deictics and frequency of occurrence in the corpus
2.1 Personal deixis
Personal deictics are of considerable interest for functional approaches to (popular)
scientific texts because they “may reveal writers’ perceptions of their own role in
research and their relationship with expected readers as well as the scientific-
academic community” (Kuo 1999: 121). The initial finding that personal deictics
are much more common in the English than in the German text endings (98 vs. 42
occurrences, cf. Table 2) is consistent with the findings of House (1997, 2006).
Summarizing her own and others’ empirical studies on English–German differences
in communicative style (see Note 12 for references), she concludes that, inter alia,
English discourse is generally characterized by an orientation towards persons
(“interactional”), while German discourse tends to be content-oriented
(“transactional”). As will be seen below, these different style preferences of English
and German are indeed the most likely cause of the observed frequency difference.
Table 3, which lists the personal deictics found in the corpus, shows that almost
exclusively plural speaker deictics (‘first person’) were encountered15, i.e. English
we, us, our and the corresponding German wir, uns, unser (which will be subsumed
under their ‘dictionary forms’ we and wir in the following; this practice will be
adopted for all deictics discussed in the remainder of this article).
Table 3. Personal deictics in the corpus
we, us, our (90)
wir, uns, unser (39)
you, your (5)
I, my (3)
ich, mein (3)
The deictics were categorized according to referent type, where a broad distinction
can be made between reader-exclusive and reader-inclusive uses (cf. Harwood
2005). The results of this first classification can be seen in Table 4.
Table 4. Inclusive and exclusive uses of English we and German wir
in direct speech
The table suggests that the frequency difference between the English and the
German text endings is due to the abundance of reader-inclusive uses of we in the
English texts. These will be given a closer look in the following section. (The 5
occurrences of we with unclear referents and the single occurrence in direct speech
will not be dealt with.)
2.1.1. Inclusive we/wir. Interestingly, the referent of almost all encountered uses of
inclusive we is as extensive as society (examples 6 and 7) or even humankind
(examples 8 and 9). That is, almost all occurrences of we in the corpus can be
classified as “rhetorical we” (Quirk et al. 1985, §6.18).
Today, as we live longer, exercise less, eat too much and smoke, many of us
suffer from inflammation’s dark side [...].
We must also become more proactive in addressing the state of our
waterways, instead of reacting to each fish kill as if it were a limited,
But because we visualize numbers as complex shapes, write them down and
perform other such functions, we process digits in a monumentally
awkward and inefficient way.
We are still some 80,000 years from the peak of the next ice age, so our first
chance for an answer is far in the future.
The term “rhetorical” is quite a fitting label for the uses of inclusive we found in the
corpus. By using the deictic, authors seem to pursue the rhetoric goal of
highlighting the relevance of their claims and findings not only to humankind in
general, but particulary to the reader of the article, who is included in the ‘global’
reference of inclusive we. The deictic thus serves authors as an important means of
achieving an addressee-oriented style. Tang and John (1999: 27) go as far as to
claim that inclusive we, “far from giving the reader information about the writer,
effectively reduces the writer to a non-entity”. The few uses of inclusive wir in the
German text endings are used in a similar way:
Auf diese Weise lernen Physiker mehr über die fundamentale Struktur
unserer Welt [...].
‘In this way, physicists learn more about the fundamental structure of our
Die Dunkelheit des Nachthimmels berichtet uns also von der zeitlichen
Endlichkeit des Kosmos [...].
‘The darkness of the night sky thus tells us of the temporal finiteness of the
Was können wir also tun, um der BSE-Krise ein Ende zu bereiten?