2.1 The distribution and semantics of Danish and English prepositions

Prepositions are typically used to cover spatial and temporal domains. Crosslinguistically, there is wide variation in which meanings are distinguished and how different meanings are distributed among different prepositions. Even prepositions in closely related language do not match up. The Danish prepositions for, i, på, and til cover similar, although distinct, semantic domains when compared to the English prepositions for, in, on, and to, as shown in Figure 1. These distinctions exist despite similarity in form and etymology. Danish i and til are cognate to English in and to. Despite being both cognates and homographs, Danish for and English for rarely overlap in use. Danish på and til have a wider range of uses than English on and to.

Figure 1. Distributional overlap for English and Danish prepositions.

Note: Proto-scenes (according to Tyler & Evans Reference Tyler and Evans2003) for the English prepositions in, on, to, and for on the left. The middle part lists selected examples from Tyler & Evans (Reference Tyler and Evans2003). TR = trajectory, LM = landmark. The dotted lines covering the English examples and the Danish prepositions to the right illustrate how each underlined preposition in the English examples translates into Danish. The asterisk in one of the examples denotes that the English example could be translated by either Danish for (if the meaning is ‘on behalf of Carol’) or til (if Carol is the recipient).

In the following, we consider the semantic distribution of the four English prepositions and their Danish counterparts one by one. We discuss the proto-scenes of each preposition taking the cognitive analysis of English prepositions proposed by Tyler & Evans (Reference Tyler and Evans2003) as the point of departure. Under this analysis, proto-scenes are defined as abstract mental idealizations over experienced spatial scenes, which can be coded by prepositions. For instance, the figure shows that a Danish translation of the pear is in the bowl would use the Danish preposition i, whereas the preposition på would be used for the translation of the cow munched grass in the field.

As shown in Figure 1, the proto-scene for English in and Danish i both cover full (in a box) and partial containment (in the bowl) The Danish preposition på shares the contact meaning with English on, but covers a wider semantic field and overlaps distributionally with English in and to. In addition (not depicted in Figure 1), på overlaps with at in expressions like at the pharmacy ‘på apoteket’.

English to and Danish til differ in proto-scene. For English to, the proto-scene involves static orientation, while the Danish proto-scene involves motion of the TR towards the LM (one of the extended meanings of English to). For static orientation, Danish tends to use other prepositions such as på and mod.

English for and Danish for share core senses pertaining to motives, intentions, and meanings (Tyler & Evans Reference Tyler and Evans2003:153–154), but the domain covered by Danish for is much smaller. English for can be used in a ‘purpose sense’, ‘intended recipient sense’, and ‘benefactive sense’. In Danish, the benefactive sense can also be realized with for, but the purpose sense and the intended recipient sense are typically realized with til.

2.2 Crosslinguistic influence in L1 and L2 writers of Danish

When L2 users express themselves in an L2, material or structures from their L1 may be retained. This phenomenon has been called interference (Weinreich Reference Weinreich1953), negative transfer (Selinker Reference Selinker1992, Ellis Reference Ellis1994, Jarvis & Odlin Reference Jarvis and Odlin2000), and crosslinguistic influence (Sharwood Smith & Kellerman Reference Sharwood Smith and Kellerman1986, Jarvis & Pavlenko Reference Jarvis and Pavlenko2008). Previous studies of crosslinguistic influence in the written production of prepositions have primarily been concerned with the use of prepositions in English L2 texts and how the use of English prepositions is influenced by the L1 of the writers (see for example Mueller Reference Mueller2011, Eddine Reference Eddine2012, and Bakken Reference Bakken2017 for studies on crosslinguistic influence from Spanish/Chinese/Korean, French, and Norwegian). Different types of evidence have been used to argue for crosslinguistic influence in the L2 (Jarvis Reference Jarvis2010), such as homogenous patterns of language use within a group of L2 users with the same L1, but heterogeneous patterns between groups of L2 users that have different L1s. Other types of evidence for crosslinguistic influence are cross-language congruity (similarity between language use in the L1 and in the L2) and intralingual contrasts (features that differ between the L1 and the L2).

However, crosslinguistic influence is not solely a characteristic of L2 language use. In multilingual language users, a second or any number of additional languages may also influence the L1 (Cook Reference Cook and Cook2003). Influence from L2 onto L1 has been observed in all areas of language structure, including phonology, morphosyntax, lexicon, and pragmatics (Pavlenko Reference Pavlenko2000, Cook Reference Cook and Cook2003). The influence may be temporary or permanent and range from borrowing transfers (such as lexical borrowings) to loss of the ability to produce certain L1 elements due to L2 influence (Pavlenko Reference Pavlenko and Cook2003). Attested examples of L2 influence on the L1 also include restructuring transfer and shifts (for example semantic extension), as well as convergence, the creation of a unitary system which is distinct from both the L1 and the L2 (Pavlenko Reference Pavlenko and Cook2003). Examples of crosslinguistic influence from L2 English on Scandinavian languages include influence on phraseology (Gottlieb Reference Gottlieb, Furiassi, Pulcini and Rodríguez González2012 on Danish) and vocabulary (Sundqvist Reference Sundqvist2009 on L1 Swedish).

In studies of crosslinguistic influence, not all types of evidence are feasible to obtain (Jarvis Reference Jarvis2010). Our study focuses on intralingual contrasts, discussing when the non-standard use of a preposition in Danish corresponds to its typical use in English or not. It is reasonable to assume that the L1 and L2 Danish texts in our study will both exhibit crosslinguistic influence from English. The L2 users have English as L1 and the L1 users are high school students who have been taught English as a foreign language since primary school. They are frequently exposed to English in Danish and foreign media. Influence from English in the written production of Danish by L1 users has been observed for individual words as well as phrases (Sørensen Reference Sørensen and Akhøj Nielsen2010, Gottlieb Reference Gottlieb, Furiassi, Pulcini and Rodríguez González2012). Additionally, young people living in Denmark tend to have positive attitudes towards loans from English (Heidemann Andersen Reference Heidemann Andersen, Widell and Kunøe2003) and may therefore be inclined to accept and produce calques from English.

Danish and English prepositions are similar in prototypical meaning but differ in extended meanings (see Section 2.1). In some cases, the cognates and partial overlap between Danish and English prepositions may facilitate production, especially in cases where the prepositions denote prototypical meanings. In other cases, calquing from English may conflict with the norms of standard Danish, for example with respect to prepositional choice and the valence of verbs (Gottlieb Reference Gottlieb2004). Yet, crosslinguistic differences and similarities cannot account for all variation in preposition usage. Previous studies of L2 users have shown that structures that differ from their L1 are not necessarily challenging, and that non-standard L2 usage also occurs in areas where the L2 structure corresponds to their L1 (Eckman Reference Eckman and Jung Song2011:620). Thus, we cannot assume a priori that they will struggle with Danish prepositions that have a different distribution than their cognates in English, and we cannot assume that L2 texts will conform to standard usage of Danish in sentences where the preposition use corresponds to equivalent English sentences. Section 2.3 is concerned with non-standard usage of valence-bound prepositions and Section 2.4 concerns prepositions with temporal meaning. Both types are mentioned by Lund (Reference Lund1997) as specific challenges to L2 learners of Danish, and we therefore explore whether the patterns of L2 users are distinct from those of L1 users. The third type of non-standard preposition usage is that of word class confusions, which we see as related to phonological similarity. There are, to our knowledge, no previous studies on confusion of prepositions with non-prepositions, but our data suggest that it is a relevant category to explore.

2.3 Crosslinguistic influence for valence-bound prepositions

In valence theory, obligatory complements are ‘those complements needed to form a grammatical sentence with the governing word (in a particular sense)’ (Herbst et al. Reference Herbst, Heath, Roe and Götz2004:xxxi). Prepositional phrases can be obligatory complements to verbs, adjectives, and nouns, and thereby part of their valence patterns. In these cases, the verb, adjective, or noun may select a specific preposition. Verbs that select a specific preposition are also known as prepositional verbs in some theories (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1972:811; Vestergaard Reference Vestergaard2019). In example (1), the verb believe selects in, while the adjective keen selects on in (2).

A valence-bound preposition cannot be exchanged with a different preposition (*believe on science). Still, the meaning of the prepositions is opaque in (1) and (2), compared to prototypical spatial uses where in clearly expresses containment (in the box) and where on clearly expresses contact/support (on the table). The English expression in (1) translates to Danish tro på and in this case, the proto-scene of the preposition på involves contact/support, rather than containment. This arbitrary crosslinguistic difference in choice of preposition illustrates the challenges for L2 users. Lund (Reference Lund1997) argues that specifically the opaque semantics of valence-bound prepositions makes them more challenging to learners of Danish as L2 than non-valence-bound prepositions whose semantics are more transparent.

We therefore examine whether valence-bound prepositions are a specific challenge to L2 users. The analysis focuses specifically on prepositions bound by verbs and adjectives since valence-bound prepositions are more frequent with verbs and adjectives than with nouns and because the valence patterns of nouns are not as well-defined.

2.4 Differences between spatial and temporal uses of prepositions

Temporal prepositions have been described as semantically less transparent than spatial uses of prepositions (Lund Reference Lund1997). From this perspective, the temporal use of prepositions is seen as a type of extended use, assuming that the domain of time is understood through the domain of space (Lakoff & Johnson Reference Lakoff and Johnson1999, Tyler & Evans Reference Tyler and Evans2003, Johansson Falck Reference Johansson Falck2014). The spatial use of the preposition in in the box clearly expresses a containment relation. In temporal uses, such as in May, the understanding of May as a container relies on a TIME is SPACE metaphor (Lakoff & Johnson Reference Lakoff and Johnson1999).

According to Lakoff & Johnson (Reference Lakoff and Johnson1999:153) a fixed duration of time can be understood as ‘a bounded region on a path along which an observer moves’ and can either be conceptualized as a container or as a point location along a line. It is, however, not entirely semantically transparent why some fixed durations of time are conceptualized as containers (in May, in the afternoon), while others are conceptualized as point locations (on Monday, at noon, at Easter).

Although related languages tend to have the same spatiotemporal metaphors, such as the TIME ORIENTATION metaphor, the MOVING TIME metaphor and the MOVING EGO metaphor, the conceptualization of temporal units differs, even between closely related Germanic languages (Lakoff & Johnson Reference Lakoff and Johnson1999, Durrell & Brée Reference Durrell, Brée and Zelinsky-Wibbelt2011, Johansson Falck Reference Johansson Falck2014). For instance, in contrast to English, the Danish sentence Hun løb 10 kilometer på 30 minutter ‘She ran 10 kilometres in 30 minutes’, employs the preposition på ‘on’, rather than i ‘in’ (for an overview of Danish prepositions employed in expressions of time compared to English, see Lundskær-Nielsen & Holmes Reference Lundskær-Nielsen and Holmes2011:165–166).

Lund (Reference Lund1997) suggests that L2 users of Danish acquire spatial uses before temporal uses, since temporal uses are more semantically opaque. Later results from empirical studies of other languages, however, have provided little support for the hypothesis that L2 users acquire spatial uses before temporal uses and instead point to the role of collocational strength, frequency, and similarity to prepositions in the L1 (Lam Reference Lam2018). One could suspect that temporal preposition phrases such as i dag ‘today’ and på mandag ‘on Monday’ would be acquired earlier simply because they are frequent (in usage and in classrooms) and often lexicalized. To further investigate this matter, our study examines whether the share of non-standard tokens in Danish L2 texts is higher for temporal uses than for spatial expressions and for non-spatiotemporal uses.

2.5 Aims and hypotheses

We carry out a comparative analysis of non-standard preposition usage in L1 and L2 texts, focusing on the differences that can be found in the use of spatial vs. temporal prepositions, valence-bound vs. non-valence-bound prepositions, and word class confusion. We expect that there is non-standard use of prepositions in both L1 and L2 texts.

In the quantitative part of the study, we also test the following two hypotheses, which are inspired by Lund’s (Reference Lund1997) acquisition hierarchies for L2 Danish.

• In the L2 texts (but not the L1 texts), prepositions bound by verbs and adjectives will have a higher share of non-standard tokens than other types of prepositions.

• In the L2 texts (but not the L1 texts), temporal uses will have a higher share of non-standard tokens than spatial expressions.

In the qualitative part of the study, we consider possible calquing from English.

3.1 Annotation of prepositions in the corpus

Each text in the corpus was proofread and annotated by two linguists at the Broken Grammar and Beyond research project at University of Copenhagen. The proofreaders annotated and categorized all occurrences of non-standard language usage (whether related to prepositions or not) using XML and suggested how the text could be corrected. The tokens that were marked up included non-standard orthography, non-standard word order, and non-standard use of morphology and lexicon. The proofreaders consulted Danish dictionaries and grammars, other L1 users as well as Google searches to identify all occurrences that did not conform to the norms of standard Danish.

Using the fully annotated corpus, we extracted all tokens that involved non-standard use of a preposition (except for those that purely involved non-standard orthography such as spelling til as ti). We extracted all occurrences of the following three subtypes: omitted prepositions, superfluous prepositions, and the confusion of two prepositions. From this set of tokens, we analyzed only those cases that involved the four target prepositions: for, i, på, and til. In addition, we extracted all tokens that involved confusion between any of these four prepositions and a non-preposition (1 token in L1, 11 tokens in L2). Examples and definitions are shown in Table 3. The full set of non-standard tokens for individual participants is shown in Appendix A in the supplementary material.

Table 3. Definitions of non-standard usage involving the prepositions på, i, for, and til.

There is not much dialectal variation in written Danish, and dialect forms are not accepted in official writing as regulated by the Danish Language Council, Sprognævnet. Yet, sociolectal and idiolectal differences in the norms of preposition can make it difficult to distinguish non-standard cases that conflict with written norms from natural variation. In cases of doubt during our annotation, we erred on the side of caution and excluded the specific case from the analysis.

4.1 Quantitative analysis

We used linear mixed models (R version 4.1.1, package lme4) to test (i) whether prepositions bound by verbs and adjectives are more prone to non-standard usage in L1 and L2 than other prepositions, and (ii) whether temporal prepositions in L1 and L2 are more prone to non-standard usage than spatial prepositions. The dataset for these two analyses included all correct uses of the four target prepositions as well as tokens where a target preposition was omitted or confused with another preposition. For each token we annotated (i) whether the preposition was bound by a verb or adjective, and (ii) whether the use of the preposition was spatial, temporal, spatiotemporal, or none of these. Word class confusions (confusions between a preposition and a non-preposition) and superfluous uses of a preposition were not included in the dataset for these analyses, since such tokens cannot meaningfully be categorized as spatiotemporal or valence-bound.

4.2 Qualitative analysis

Throughout the qualitative analysis, we discuss whether crosslinguistic influence from English can be observed in the form of possible calques. A calque (or loan translation) can be defined as ‘a complex lexical unit (either a single word or a fixed phrasal expression) that was created by an item-by-item translation of the (complex) source unit’ (Haspelmath Reference Haspelmath, Haspelmath and Tadmor2009:39). The source unit may be lexical or morphosyntactic. In our study, sentences that are possible English calques are thus non-standard Danish sentences that look like near word-for-word translations (or preposition-to-preposition translations) of an equivalent English sentence as illustrated by example (3). Here the Danish preposition i is used instead of the target på in a context where the correct choice of preposition in English would be in.Footnote ¹

In other cases, there are no indications of calquing. For instance, if we backtranslate the non-standard sentence in (4) to English from Danish, the result is not a grammatical sentence in English either (we shall meet *to my home).

In the English backtranslation, the correct preposition would be in, and in Danish its cognate preposition is i. In this case, it is therefore implausible that the use of til is due to crosslinguistic influence from English.

In some cases it is difficult to determine whether or not there are possible calques. Consider example (5), for instance.

In our analysis, we classified til as a superfluous preposition since the target form ‘Kan du besøge os?’ does not have a preposition. As indicated by the translation, we interpret the semantically equivalent English expression of the target form to be ‘visit us’. This translation implies that the sentence is not a possible calque. However, one could alternatively assume that the structurally equivalent expression would be pay a visit to us. By this analysis, the Danish sentence would express crosslinguistic influence from English.