4.1. Code-Switching

Of the 186 instances of transferred verbs in the main dataset, 136 are categorized as code-switches because they involve the use of English-origin verbs.Footnote ²⁷ This figure of slightly over 73% suggests that code-switching is the most common strategy for integrating English verbs into TxG.Footnote ²⁸ Here, I discuss the semantic domains of code-switched verbs and then classify them according to the type of morphological marking they display.

A total of 79 different verbs from a wide range of semantic domains occur among the 136 code-switching instances. Table 2 provides an overview of the code-switched verbs (in their English forms) organized according to semantic domains.

Many verbs come from domains identified in previous research as prominent sources of code-switches, such as administration, agriculture, religion, and transportation. Several other code-switches name concepts that did not feature in 19th-century European culture or were integral to the rural Texan lifestyle. Boas & Pierce (Reference Boas and Pierce2011) also noted the domains of education and transportation as prominent sources of code-switches, but only four verbs in the dataset come from these domains, namely, graduate, drive, pave, and shift [gears].Footnote ²⁹ Another source of multiple code-switches, including the verbs join and station, is the domain of military service, which was not identified in previous research.

However, the exact range of domains of code-switched verbs does not entirely overlap with the code-switching domains identified in previous studies, which focus primarily on code-switched nouns. Interestingly, over half of the code-switched verb types (45 of 79) are not expected based on previous research, as they do not pertain to prominent domains for code-switching (such as agriculture or administration); or they refer to concepts not novel to U.S. or Texan life and for which the donor dialect would have its own verb. These are listed in the final two rows of table 2. Most of these verbs have fairly rich semantics (for example, behave, believe, deprive, fight, rotate, separate), but a handful of them are highly general verbs (for example, do, go, make, like, run).

There are several possible explanations for the use of English verbs by TxG speakers and for the lack of overlap between the semantic domains of code-switched verbs and the domains identified in previous research. Of course, the verbs in the top rows of table 2 are likely code-switched because the original immigrants did not have verbs to express these concepts. In the case of the verbs in the “Other” row, one probable explanation for code-switching is that speakers had never (or only rarely) been exposed to the StG counterparts of those verbs. This is likely the case with infrequent verbs such as deprive, sterilize, or volunteer.

Another potential reason for TxG speakers to code-switch English verbs could relate to differences in the argument structure of English and German verbs. For instance, while English agree only requires a prepositional object in a with PP, German equivalents such as zustimmen, einigen, and übereinstimmen further require reflexive direct objects and/or additional prepositional phrases. Another such verb is English retire, which can be used in simple intransitive constructions (She retired) but has several German equivalents, which involve several words, such as in Rente gehen ‘go into retirement’, sich pensionieren lassen ‘let oneself be pensioned’, or in den Ruhestand treten ‘step into retirement’. While these structural differences may not necessarily trigger code-switching in fluent bilinguals, TxG speakers may have a weaker command of these complex structures because of language attrition (or because they might have never learned them in the first place). These verbs underscore the complex nature of integrating foreign verbs into native language structures: Verbs—unlike nouns—are an integral part of their argument structure construction, and they cannot simply be inserted into foreign structures, as is often the case with code-switched nouns.

The code-switching of more general verbs such as be, make, and run (bottom row of table 2), however, cannot be attributed to attrition or structural differences, as these verbs are highly frequent and do not involve complex structures. Indeed, these code-switches are quite infrequent in the dataset, with each verb occurring only one time except for make (3 instances) and like (21 instances, discussed below). One potential explanation is that these verbs are borrowed as parts of idioms or English-specific senses.Footnote ³⁰ Their usage in TxG may result from the simultaneous activation of multiple linguistic systems in speech production (see Goldstein Reference Goldstein2004), during which speakers utter the more easily accessible verb: Because the speakers activate a multi-word expression based on an English idiom (that is, a grammatical construction), they adopt the English verb as part of the expression. In sum, rather than attributing the use of English verbs to attrition or unfamiliarity, one must also recognize that structural complexity and idiomatic collocations may determine whether a verb is code-switched or not.Footnote ³¹

The analysis of code-switched verbs must also go beyond a mere lexical and semantic description because German morphological marking for tense, person, number, and so forth differs from that of English. Here, I investigate how English verbs are morphologically integrated into TxG. The phonological similarity of several affixes in German and English makes it very challenging to determine which language provides the morphology.Footnote ³² Thirteen examples could not clearly be assigned German or English morphology due to such overlap between the German and English verbal affixation systems. Another eight examples were omitted from the morphology analysis due to background noise or other issues in the original recordings. The remainder of this section therefore describes the affixation patterns of the remaining 115 code-switches. The examples fall into four main categories: German morphology, partial German morphology, English morphology, or combined English and German morphology. Each category is discussed in turn below.

The most common strategy involves the use of an English root and one or more German morphological markers. Thirty-four of the 115 analyzable examples exhibit StG morphological marking, while 46 examples exhibit German marking with slight deviations from the standard variety. Among the 34 purely German affixed examples, 22 have the final -en affix, which marks infinitives and 1st/3rd person plural forms, as in 2a and 2b, respectively.

1. (2)

   
   Footnote ³³

Seven examples exhibit the ge- prefix and a -t suffix, as in 3a, and one an -en suffix, as in 3b.Footnote ³⁴

1. (3)

   

Finally, four examples exhibit only a suffix clearly pronounced as German [t]. Three such examples mark past tense or past participle, while one indicates a 3rd person singular present tense, as in 4.

1. (4)

   

I now turn to the 46 examples exhibiting partial German morphology that differs slightly from that expected in StG. Ten such examples exhibit the ge- prefix, but do not have any suffixes, as in 5.

1. (5)

   

The most prevalent pattern, both in this category and in the entire main dataset, are verbs with the German ge- prefix and suffixes that are indistinguishable between German -(e)t and English -(e)d. Thirty-six such examples were found in the dataset. The majority of such examples involve a root that ends in a voiceless consonant and would thus require the [t] allomorph in English, as in the following example:

1. (6)

   

Because these examples follow the affixation patterns virtually identical to those expected for StG verbs, it is unlikely that they have been influenced by English morphology.

The next major category is comprised of 22 examples. The English-origin TxG verbs in this category carry morphological marking closely related to that of the English-origin verbs. In 13 instances, the verbs exhibit no (ge-) prefix and a voiced -(e)d suffix, as in paved in 7, which makes this the most frequent pattern.

1. (7)

   

Seven of the examples that carry English morphology do not involve any affixes; they correspond to English forms that do not require tense-marking affixes. Five such unaffixed forms are used in infinitival contexts, as in 8a, while two unaffixed forms are used for plural present tense forms, as in 8b, both cases in which StG verbs require -en suffixes.

1. (8)

   

Two other examples clearly exhibit English morphology, in that the verb form undergoes ablaut, as expected in English (got in 9).Footnote ³⁵

1. (9)

   

The final type involves a combination of English and German morphological marking. Most prevalent among these cases (11 examples in the main dataset) are verb forms exhibiting the German ge- prefix and a suffix, which is clearly voiced as -e(d), as expected in English:Footnote ³⁶

1. (10)

   

The two remaining examples exhibit the ge- prefix with an ablauted form of the verb, as expected in English. In 11, for example, the verb drive has an English past participle form with a German prefix.

1. (11)

   

Table 3 summarizes the distribution of morphological marking types in the main dataset.

Most code-switched verbs (80 of the 115 analyzed examples; 70%) exhibit some aspects of German morphological marking, as expected based on previous research discussed in section 2. Thirty-four examples clearly exhibit German marking, while 46 exhibit partial German marking or German marking that is partially ambiguous. Twenty-two examples (19%) exhibit marking identical to that expected in English. Thirteen examples (11%) involve some aspects of both German and English morphology. Finally, 21 examples were omitted from analysis due to ambiguity between German and English marking or to technical issues (for example, unclear audio). These findings largely corroborate the results of earlier research on TxG and those of Wichmann & Wohlgemuth (Reference Wichmann, Wohlgemuth, Stolz, Bakker and Salas Palomo2008), who claim that English verbs borrowed into (other) Germanic languages follow the direct insertion strategy, whereby the foreign verb is directly inserted into recipient language structures.Footnote ³⁷ However, the existence of forms that do not follow the expected TxG morphology suggests that the direct insertion strategy does not fully capture all instances of code-switched verbs in TxG.

So far, it has been shown that code-switches make up the majority of instances of English verb transfers in the main dataset. Nearly half of the verb types come from domains that reflect cultural differences between Texas and 19th-century Europe—such as administration, agriculture, religion, new technologies, and rural life. The existence of such domains is expected based on previous studies on TxG code-switched nouns. However, there are just as many verb types that belong to more general semantic domains or express concepts that existed in 19th-century Europe. With respect to their formal integration, code-switched verbs can be subclassified according to the type of morphology assigned. In most cases, they are integrated into the German morphological structure, but they may also occur with English morphological marking, or a combination of both English and German marking.

4.2. Lexical Borrowing

I now discuss B&D’s category of lexical borrowing. This category includes synchronic code-switches that become conventionalized as lexical items of the recipient language. The difficulty of distinguishing nonce code-switches from established borrowings is well-documented in the literature (Haugen Reference Haugen1950, Backus Reference Backus, Isurin, Winford and De Bot2009).Footnote ³⁸ Ideally, one may look at data from earlier stages of the language to determine whether transferred words from those stages are also prominent in synchronic data. Here, Jordan’s (Reference Jordan1977) list of frequent English-origin words is compared with the present-day TGDP data to determine whether the verbs on that list are frequently used today. Another strategy is to compare the frequencies of code-switched words with their native counterparts under the assumption that a high relative frequency of a foreign word suggests that this word is established as a borrowing. This strategy is employed when dealing with the three most frequent verbs in the main dataset: retire, like, and move (change residence).Footnote ³⁹

Jordan’s (Reference Jordan1977) list of English-origin verbs in TxG includes 63 verbs. The entire TGDP corpus was searched to determine whether the verbs on Jordan’s list also occur frequently in the more recent TGDP data, which would suggest that they have become established borrowings in the TxG lexicon. Only three verbs from Jordan’s list appear with high frequency in the TGDP corpus. These verbs are farmen ‘farm’ (27/23), raisen ‘raise (livestock)’ (20/15), and (auf)picken ‘pick (up)’ (19/15).Footnote ⁴⁰ Six other verbs appear relatively frequently in the TGDP corpus (four to nine utterances in total uttered by two to six different speakers): (auf)ringen ‘ring up’ (9/5), changen ‘change’ (6/6), fixen ‘repair’ or ‘prepare (a meal)’ (7/2), phonen ‘phone’ (4/4), shucken ‘shuck (corn)’ (7/6), smoken ‘smoke meat’, (6/4), kicken ‘kick’ (4/3), killen ‘kill’ (4/3), and fighten ‘fight’ (3/3).Footnote ⁴¹ Seventeen other verbs appeared less frequently, uttered only once or twice by one or two speakers.Footnote ⁴² Finally, 33 verbs identified as borrowings by Jordan (Reference Jordan1977) did not appear in the TGDP corpus.Footnote ⁴³

As with the code-switches from the main dataset, most verbs that have likely become established borrowings refer to concepts prominent in 20th- (and 21st-) century Texan society but not in 19th-century Europe, such as (auf)ringen, phonen, shucken, and smoken. However, other borrowed verbs—such as changen, fixen, kicken, killen, and fighten—are not so easily explained, as they express concepts that were likely as prominent in the original settlers’ culture as they are today. In fact, of the 11 most frequent verbs out of 136 code-switched verbs in the main TGDP dataset, only two (shucken [corn] and farmen) were identified by Jordan (Reference Jordan1977). Among these 11 verbs, six verbs were code-switched by two different speakers (graduate, shuck [corn], farm, join, separate, suppose), and two verbs were code-switched by three different speakers (rent, run). In general, this comparison of TGDP data against historical data from Jordan Reference Jordan1977 shows that although several verbs mentioned by Jordan are used frequently today and thus have likely established themselves in the TxG lexicon, many other verbs used nearly 40 years ago are no longer prominent in present-day TxG.

Returning to the main dataset of 136 code-switches, three verbs are highly frequent and may be established as borrowings in TxG: retire is used five times by five different speakers, move (change residence) is used 11 times by six speakers, and like is used 21 times by six speakers. Here, I compare the frequencies of these verbs (exemplified in 12) with that of their German counterparts on the assumption that a high relative frequency of the English-origin verb suggests that it is an established borrowing (see B&D, p. 77).

1. (12)

   

For each of these verbs, the entire TGDP corpus was searched for both the code-switched English verb and any possible German-origin or loan-translated equivalents (for example, gleichen for ‘(to) like’).

Table 4 shows the frequency of English retire in comparison with two German equivalents. Note that the English verb is frequently used as a past participle with an adjectival function, as in He is retired. The same is true for the German equivalent pensionieren, which often occurs in adjectival form as pensioniert. Other common German equivalents are the nouns Rente ‘retirement’, as in in die/der Rente gehen/sein ‘go/be into/in retirement’. These nonverb equivalents are nonetheless taken into account, as they express the same concept as the frequently code-switched verb retire.

Table 4 shows that the English verb retire is much more frequent than expressions used for similar concepts in StG, occurring over five times more frequently than any possible equivalent. This high relative frequency thus suggests that retire is an established borrowing in TxG.

Table 5 shows the frequency of code-switched like compared with German equivalents gefallen ‘please’ and mögen ‘(to) like’ and the related loan-translated verb gleichen ‘(to) like’ (loan-translated from English like, discussed in more detail below). Gleichen and mögen exhibit the same argument structure pattern as English like, with the agent appearing as subject and the liked entity as direct (accusative) object, whereas gefallen realizes the liked entity as subject and the entity who likes it as a dative object.

The data show that the verb like is more frequent than any of its (German or loan-translated) equivalents. However, its relative frequency over its counterparts is not as drastic as that of retire, as gefallen and gleichen also occur regularly in the data. Therefore, while like is likely an established borrowing for many TxG speakers, it still competes with native German forms and the loan-translated gleichen.Footnote ⁴⁴

Table 6 shows the number of occurrences of English code-switched move (change residence) compared with German equivalents containing the root ziehen. Similar to English move, which appears with particles as in move in, move out, or move around, German ziehen occurs with various prefixes and prepositions to specify different aspects of the moving process.

A code-switched form of English move occurs 68 times in the data, but various forms of the German verb ziehen ‘move’, such as einziehen ‘move in’, nach X ziehen ‘move to X’, or hierherziehen ‘move here’, occur almost as frequently as move, with 60 instances in the corpus. Therefore, although English move is frequent, the stability of its StG counterpart(s) in TxG makes it difficult to classify as a fully established borrowing. It also remains to be determined whether the English verb is used variably by different individuals or speaker communities, or whether it is in free variation with its German counterpart(s).

This comparison of English code-switches with their German counterparts, based on relative frequency data in the TGDP corpus, shows that retire is an established lexical borrowing, like is fairly well established but still competes with various German counterparts, and move is also well established but still competes (more fiercely) with ziehen, whose various forms appear equally frequently in the corpus. There are several potential reasons for the borrowing of these verbs. Move and retire express concepts that were not as relevant in 19th-century Europe as they are nowadays: The original settlers were much less mobile than Texas Germans are today, and there was not an established system for retirement. However, a cultural explanation does not hold for like. Another reason may involve the argument structure of German equivalents, as described above: The complexity of the German patterns of argument realization could have encouraged speakers to employ an English item with simpler grammatical requirements. Again, this explanation may hold for move and retire, but the German verb mögen exhibits a similar argument structure to English like. Thus, while cultural and structural factors may (partially) explain the borrowing of rent and move, no plausible explanation for the frequent code-switching of like is available at present.

4.3. Loan Translation

As discussed in section 3, loan translation involves translating an English word using a formally related German word that does not have the same meaning in the original dialect. Repeated use of the loan-translated word with this meaning leads to the (diachronic) process of lexical change, whereby the meaning of the German translation is extended to include the meaning of the formally related English original (see section 4.5). Loan translations occur less frequently than code-switches in the dataset, as 37 loan translations were found among 186 total transferred verb instances (20%).Footnote ⁴⁵ The analysis in the remainder of this section draws on B&D’s subclassification of loan translations based on the type and number of morphemes involved; it also builds on their finding that loan translation may lead to structural interference. My analysis facilitates a comparison between loan-translated and code-switched verbs and enables one to assess B&D’s specific claims presented at the end of section 3.

With respect to loan translations of content words, B&D (pp. 82f.) propose different categories for single-word, two-word, and multi-word loan translations. Examples 13–15 contain what seem to be one-word loan translations. In 13, the English verb like ‘be fond of’ is loan-translated into TxG as gleichen, which can only mean ‘be similar to’ in StG.

1. (13)

   

B&D (p. 83) refer to this type of loan translation as semantic extension, as the original meaning of the verb ‘be similar to’ is supplemented with a new meaning ‘(to) like’ based on its phonological similarity to the English verb.

In 14, the English expression to make it (to a place) is translated using the German verb machen ‘make’, even though StG does not use machen in this sense.

1. (14)

   

Although this example also involves one loan-translated verb, it differs somewhat from the example in 13 with gleichen: The verb machen is modeled on a more complex English idiomatic expression, which requires not only the verb but also a directional prepositional comple-ment. Here, only the verb is loan-translated, but the remaining portions of the construction are expressed using TxG.

In 15, the verb nehmen ‘take’ is used in the sense of ‘take an exam’, whereas StG employs the verb schreiben ‘write’ in this context.

1. (15)

   

In the case of 15, it is possible that the loan translation of the verb is triggered by the code-switching of the preceding noun exams, which in English requires the verb take. While the most appropriate category in B&D’s taxonomy for examples 14 and 15 is one-word loan translations, these examples differ from the semantic extension of gleichen in 13, as they are used as parts of larger collocations. In fact, B&D (pp. 91–92) claim that loan translations often occur as parts of larger fixed expres-sions, particularly object-verb or prepositional phrase-verb collocations.

There are several other cases, particularly those that involve complex verbs with prefixes or particles. These cases cannot be clearly classified as one- or two-word loan translations due to structural and orthographic differences between German and English. In 16, the speaker wishes to express that certain families in the area housed children. The speaker translates the English expression ‘take in (for example, guests)’ directly into German as reinnehmen (participle reingenommen), an expression not found in StG.

1. (16)

   

Similarly, 17 is based on the complex English expression ‘set somebody up in something’, in the sense of assisting them and providing them with opportunities. When talking about a man who received land and ‘set his sons up’ in farming, the speaker translates this expression into TxG as aufsetzen, which is not used in this sense in StG; rather it is used as ‘put on’ or ‘attach’.

1. (17)

   

Orthographic differences between English and German muddle the classification of this example according to the number of words involved. On the one hand, the lexico-grammatical rules of StG require the particle and verb root of participle forms to occur together, suggesting that it is a one-word loan translation. On the other hand, equivalent English expressions are written with the particle separate from the verb, which suggests that they should be classified as two-word loan translations. Such data show that it is not always trivial to categorize loan translations based on the number of morphemes involved.

An example of a two-word loan translation is given in 18, where ‘make a living’ is directly translated as Leben machen. This expression does not exist in StG; instead, a similar meaning would be expressed using Auskommen haben lit. ‘have subsistence’ or Leben bestreiten lit. ‘contest life’.

1. (18)

   

Example 18 is quite similar to the Exams nehmen example in 15 above, as both of these utterances involve the loan translation of a verb within a specific verb-object collocation. They differ only in that the object in 15 (Exams) is code-switched from English, while the object in 18 (Leben) is loan-translated. Once again, these two examples show the difficulty of classifying loan translations based on the number of morphemes involved.

Multi-word loan translations consisting of more than two content morphemes are slightly less frequent in the main dataset.Footnote ⁴⁶ Indeed, many of these loan translations involve the translation of just one morpheme, but with repercussions for the surrounding words. One candidate for such a loan translation is provided in 19, which involves the use of German machen ‘make’ in the sense of ‘cause someone to do something’. While only the verb machen is loan-translated, it is likely that the speaker is translating the entire construction ‘it made me think’ along with machen.

1. (19)

   

A similar situation holds for 20, which contains the loan translation of the phrase ‘take up someone’s time’. As in the previous example, this utterance involves only the translation of one complex verb (or a verb-particle pair), but it is modeled on a multi-word English expression.

1. (20)

   

These examples show that B&D’s classification of content word loan translations into one-, two-, and multi-word categories is not as clear-cut as it appears. Specifically, while these examples contain only one (or two) clearly loan-translated words, the speakers are structuring the entire utterance based on complex grammatical constructions rather than on individual words.

B&D’s next major category of loan-translated items covers functional morphemes. One type of functional morpheme that seems prone to loan translation in TxG is argument-marking preposition. In these cases, TxG speakers use a German preposition with the same (or similar) phonological form as the English preposition (assigned by the verb), and not the preposition used with the same verb in StG. In 21, for instance, the speaker uses the preposition in analogously to the English construction ‘be interested in’, rather than an, which is used in the StG construction interessiert sein an ‘be interested in’.

1. (21)

   

Similarly, in 22, the speaker uses the preposition für ‘for’ on the basis of English ‘wait for’, rather than the expected preposition auf.Footnote ⁴⁷

1. (22)

   

Another type of functional morpheme susceptible to loan translation is the auxiliary in perfect tense constructions. StG employs either haben ‘have’ or sein ‘be’ to support past participles, depending on whether the verb expresses motion toward something or a change in state (sein) or not (haben), with some exceptions. English, however, only uses the auxiliary have to introduce participles, except in some fossilized collocations (for example, He is risen). In 23, the verb gehen ‘go’ (participle gegangen) is used with the auxiliary haben, rather than the expected sein.

1. (23)

   
   Footnote ⁴⁸

While the above examples of code-switching and loan translation could largely be subsumed within the categories proposed by B&D, I now turn to another type of language contact phenomenon that requires an extension of B&D’s classification.

4.4. Loan Translation/Code-Switching Hybrids

One type of phenomenon that does not clearly fit into any of the relevant categories proposed by B&D are verbal constructions involving both code-switching and loan translation. Thirteen such instances were found in the dataset, comprising nearly 7% of all examples. Each of these cases involves a complex particle or prefix verb, with either the root or the particle/prefix being code-switched and the other element being loan-translated. Eleven of the 13 hybrid examples involve a code-switched root and a loan-translated prefix/particle. In 24, for instance, the speaker imports the English phrasal verb ‘look up (information)’, but loan-translates only the English particle up with the German verbal prefix auf (rather than StG nach), while using the German verb root gucken ‘look’.

1. (24)

   

Similarly, 25 is modeled on English ‘jack up (a car)’. Here, the speaker code-switches the English verb jack and uses the German prefix hoch ‘high/up’.

1. (25)

   

Less frequently (in only two of the 13 instances), the main verb is of German origin and the prefix/particle is code-switched from English. This is shown in 26, where the German main verb participle geschossen ‘shot’ is combined with the English particle off.Footnote ⁴⁹

1. (26)

   

These types of combinations of loan-translation and code-switching have been identified in previous research. Haugen (Reference Haugen1950, Reference Haugen1953) uses the term loan blends for these cases, while Johanson (Reference Johanson1998) refers to them as mixed copies. Such hybrids appear at the intersection of B&D’s categories, as they involve the transfer of both, words (code-switching) and meanings (loan translation). To accommodate these cases, B&D’s model must be expanded to include the combination of multiple transfer types within the same utterance (and even within the same word).

4.5. Lexical Change

B&D’s (p. 77) category of lexical change involves the diachronic establishment of loan translations, whereby a native lexical item becomes conventionally associated with a foreign meaning. As discussed in section 4.2, nonce loan translations may be distinguished from those that have undergone lexical change (that is, they have become established within the speech community) by means of relative frequency: If a loan-translated expression is used more frequently with the foreign-motivated meaning than with its native meaning, or if it is used more frequently than a native expression with an equivalent meaning, then such loan-translated expression has likely been established as a lexical change. The term lexical change thus emphasizes that the range of a word’s possible functions has expanded (or changed altogether) in the contact variety (for example, TxG) relative to noncontact varieties of the language (for example, StG). Potential candidates for lexical change in the TxG data include gleichen, which is frequently used to express the non-native meaning ‘like/be fond of’, as well as a group of verbs that are highly frequent and have very general and broad meanings, such as machen ‘make’ and nehmen ‘take’, as described in the following paragraphs.

Here, the relative frequency of loan-translated gleichen ‘(to) like’ is compared against its use in the StG sense ‘be similar to’, following the method employed in section 4.2. A search for gleich in the TGDP concordancer revealed 11 instances of the loan-translated meaning compared with zero instances of the StG meaning ‘be similar to’. These results suggest that gleichen has changed its meaning from ‘be similar to’ in the original dialects to ‘(to) like’ in TxG.Footnote ⁵⁰ However, such an investigation is complicated because in StG, the form gleich also has an adverbial meaning of ‘soon’ or ‘immediately’, which is used in TxG, as shown in 27. Such adverbial uses of gleich greatly outnumber the loan-translated usages of the verb gleichen (66 to 11).

1. (27)

   

At present, therefore, it appears that the verb gleichen has undergone lexical change from StG ‘be similar to’ to the English-based meaning of ‘(to) like’. However, as a morpheme, gleich still retains the adverbial meanings it had in the original dialect. As mentioned in the preceding section, the loan translation of gleichen is likely due to the fact that its form and meaning overlap with those of English (to) like.

The loan translation data also include many semantically light verbs, which are often loan-translated as parts of larger idiomatic collocations or light verb constructions; they can also acquire more structural functions. Two such loan translations are the verbs machen ‘make’ and nehmen ‘take’. TxG speakers use the verb machen in contexts where English make is expected, but not StG machen. This was shown in 19, where machen is used in the sense of ‘make somebody do something’; in 18, in the collocation Leben machen ‘make a living’, and in 14, where it is used in the sense of ‘make it to a place’. The English verb take is often loan-translated as nehmen, as in the collocation Exams nemmen in 15, where StG requires schreiben ‘write exams’, and in the collocation ‘take up somebody’s time’ in 20. Nehmen is also used in the more general temporal sense of ‘take (for example, one week of) time’, as in the expression das nemmt beinah ein Jahr ‘that takes nearly a year’ (1-78-1-5-a). The preponderance of light verbs among loan translations in the main dataset is indeed striking: Over half of the loan translation examples contain the verb nehmen ‘take’, machen ‘make’, gehen ‘go’, tun ‘do’, or laufen ‘run’. Even semantically richer verbs that are loan-translated do not have as rich of a meaning as (the vast majority) of TxG code-switched verbs listed in table 2 above. These include setzen ‘(to) set’, ziehen ‘draw/pull’, gucken ‘see/watch’, and halten ‘hold’.Footnote ⁵¹

The high frequency of loan-translated semantically general verbs is also observed by B&D (p. 92). They claim that basic vocabulary items are more likely to be loan-translated, while items with richer meanings are more frequently code-switched. B&D propose that this comple-mentarity may arise because “basic vocabulary patterns with functional elements are being produced without much conscious attention, while specific vocabulary is consciously selected.” To reframe this proposal in a (Diasystemic) Construction Grammar view of the mental lexicon, words with rich semantics (such as the code-switched verbs behave, deprive, or sterilize) are typically not elements of larger idiomatic expressions or collocations and can thus be directly code-switched from English. In contrast, the semantically light verbs appear in larger constructions and are more frequently loan-translated while other words in the constructions are uttered in German.

From a diachronic perspective, such light verbs may have undergone lexical change in TxG and thus have a broader (or different) range of functions than in the original dialect. Specifically, they may now be associated with new functions modeled on English collocations and idioms, while also maintaining the original StG functions. The close relation between German and English may encourage such lexical changes, as many idiomatic uses of these verbs are the same in noncontact varieties of English and German. However, diachronic developments have led to divergences in the exact range of functions the verbs can fulfill in each language.

In summary, the analysis of loan translations has shown that, although they are not as frequent in TxG as direct code-switches, they are more frequent than previous research suggests. As with the Turkish-Dutch data discussed by B&D, in TxG content, functional, or grammatical morphemes may be loan translated. B&D propose a subclassification of content word loan translations according to the number of words involved. However, the data revealed several examples that were difficult to categorize as either one- or two-word loan translations, particularly, particle and prefix verbs. There are also instances of code-switching/loan translation hybrids, identified by Haugen (Reference Haugen1950, Reference Haugen1953) as loan blends and by Johanson (Reference Johanson1998) as mixed copies, which do not fall neatly into B&D’s classification and require an extension of their taxonomy. The analysis also revealed three types of grammatical and functional morphemes that are susceptible to loan translation in German-English contact situations, namely, the prepo-sitions introducing prepositional objects, particles and prefixes of complex verbs, and auxiliaries in perfect constructions. These pheno-mena bleed into the realm of structural interference, as they involve reinterpretations of grammatically relevant morphemes, a phenomenon discussed in more detail below. Finally, the TxG data support B&D’s (p. 91) claim that loan-translated words are more semantically general than code-switched words, which likely relates to the more frequent occurrence of general words/verbs in idiomatic expressions or other complex constructions.

4.6. Structural Interference

I now turn to the next category in B&D’s taxonomy: structural interference. As this study focuses primarily on the lexical category of verbs, I only discuss cases of structural divergence from StG related to the transfer of verbs in the main dataset.Footnote ⁵² The discussion is also limited to instances of structural interference not mentioned in the previous subsections (prepositions, particles/prefixes, auxiliaries), namely, verbal inflection, case assignment, and word order. The structure of 28 differs from the corresponding StG structure with respect to each of these structural properties.

1. (28)

   

First, the StG participle of helfen ‘help’ is geholfen, but the verb form in this example is much more closely related to the English past (participle) form helped. Second, StG word order requires the inflected verb to appear in the second position in main clauses and in the final position in subordinate clauses; infinitives and participles appear in the final position in main clauses and preceding the main verb in subordinate clauses. In 28, however, the main verb appears directly after the auxiliary and before the object, an ordering expected in English.Footnote ⁵³ Furthermore, the object of helfen in StG is realized in the dative case, but here it carries accusative case marking (that is, die Kirche). While this latter contrast may be due to influence from English, it likely results from the more general case syncretism in TxG, whereby the dative case is in decline (Boas Reference Boas2009, chapter 5; see section 2).

Another instance of structural interference is found in examples in which German besser is used in the modal sense associated with English besser, as in better do something. In 29, the loan translation besser appears in second position, but the main verb appears in third position directly after German besser. This type of word order is not expected in StG.

1. (29)

   
   Footnote ⁵⁴

These data clearly support B&D’s (p. 79) claim that items involved in loan translation, particularly multi-word expressions, may cause structural interference. In fact, it is more likely that the speaker is transferring an entire grammatical construction from English, including both the fixed lexical item besser, and the schematic slots for the main verb and its related elements. These cases thus emphasize that contact phenomena cannot be simply reduced to the transfer of words, structures, or meanings, but require a more holistic view of linguistic constructions.

The structure of TxG clauses can also be affected by code-switching. In 30, the speaker uses the English verb supposed, in the sense of ‘be expected/supposed to’.

1. (30)

   

Here, the speaker uses a syntactic configuration that exists in StG but is not used with the StG equivalent of the code-switched verb suppose (erwartet ‘expected’). Specifically, StG expressions such as dabei sein, etwas zu tun ‘be doing something currently’ or bereit sein, etwas zu tun ‘be prepared to do something’ have the same structure as 30: The agent is subject of a predicative construction, and the activity is expressed in a final infinitival clause. However, to express the concept in 30, StG would use the participle of erwarten ‘expected’ with the dummy subject es ‘it’ and express the agent as the subject of a subordinate clause headed by dass: Es wurde erwartet, dass wir kein Englisch sprechen ‘It was expected that we speak no English’.Footnote ⁵⁵ While B&D (p. 81) point out that loan translation may result in structural interference, this example shows that even code-switching may influence the structure of the recipient language.

This cursory analysis of structural interference shows that the use of English verbs and verbal idioms in TxG influences TxG grammatical structures to some degree. Potential areas of structural transfer include word order, argument structure (for example, case assignment), inflectional patterns, as well as the argument-marking prepositions, verbal prefixes and particles, and auxiliaries of perfect constructions discussed above. My analysis shows how B&D’s taxonomy can be refined to account for specific grammatical features of German and English that are susceptible to structural interference in these types of contact situations. A richer taxonomy would also help account for how structural interference results from loan translation and code-switching.