Passive Sentences and Abstract Agents in German School Books

2.1 Passive sentences

Genus verbi is a grammatical category, which indicates the relation of the verb to the subject and to the object. A distinction can be made between active and passive. The English passive is built by a conjugated form of be and past participle and this construction can be used in different tenses, as the examples in (1) show.

(1) a. Simple present: The house is built.

b. Simple past: The house was built.

c. Present perfect simple: The house has been built.

d. Past perfect simple: The house had been built.

e. will-future: The house will be built.

The prototypical German passive is built differently: The prototypical construction does not use the verb sein, which corresponds with the English be; it uses a conjugated form of werden and participle II. For the conjugation of different tempus and genus verbi in German language, not only the auxiliary verbs sein (Eng.: be) and haben (Eng.: have) are used, but werden as well. The contexts of this verb are illustrated in the examples in (2): The German verb werden can be used for sentences in future I (2a), in future II (2b), in copula constructions (2d) and in the werden-passive (2c). According to that, the German verb werden is used in contexts, where the English language uses either the verb will in tenses, the verb become in copula constructions or the conjugated form of be in passive. Thus, a word-by-word translation from German to English and vice versa is not possible and the different use of auxiliary verbs might accompany a different use of this construction in the languages.

(2) a. future I: Tom wird das Haus bauen.

Tom will the house build.

´Tom will build the house.´

b. future II: Tom wird das Haus gebaut haben.

Tom will the house built have.

´Tom will have built the house.´

c. werden-passive: Das Haus wird (von Tom) gebaut.

The house is (by Tom) built.

´The house is built (by Tom).´

d. copula: Tom wird Lehrer.

Tom becomes teacher.

´Tom becomes a teacher.´

Active sentences and werden-passive sentences in German language are coding the same semantic roles but in different functions (Eisenberg 2013: 118–119), which can be illustrated by figure 2.

The semantic agent (Tom) takes the syntactic function of the subject and the semantic patient (das Haus) fulfils the function of the direct object in an active sentence. In a werden-passive sentence the allocation differs: The semantic patient (das Haus) has the syntactic function of the subject and the semantic agent (Tom) can be realised in form of a prepositional complement, which is optional.

The sentences in figure 1 are equivalent in the content but are characterised by different point of views: Because the agent is in the position of the facultative prepositional complement, the verb of the werden-passive is central and the action itself is focussed if the agent is not realised in the sentence ​​(Pape-Müller 1980: 124; Eisenberg 2013: 128). The reduction of the agent allows a factual representation of what is happening (Gang 1997: 30) and the prevention of redundancies in a text (Pape-Müller 1980: 234; Zifonun et al. 1997: 1841).

If the agent instead is reconstructed in form of the prepositional complement (von Tom), which is optional, this complex element moves into the focus of the statement and the representation is even focussed on the agent (Helbig 1989: 216; Dürscheid 1997: 245; Eisenberg 2013: 129). Therefore, the werden-passive can be used to present occurrences conspicuously compared with active: either more focused on the process by reducing the agent or more focused on the agent by reconstructing it.

German passive sentences can be subdivided in werden-passive, sein-passive and bekommen-passiv (Heidolph et al. 1981: 542 f.; Helbig and Buscha 2013: 143), which is illustrated in (3). The sentence in (3) a. is a werden-passive and equivalent with (2) c. and the example in figure 1. The sentence in (3) b. can be categorised as sein-passive and (3) c. as bekommen-passive.

(3) a. werden-passive: Das Haus wird (von Tom) gebaut.

The house is (by Tom) built.

´The house is built (by Tom).´

b. sein-passive: Das Haus ist (von Tom gebaut.

The house was (by Tom) built.

´The house was built (by Tom).´

c. bekommen-passive: Der Freund bekommt ein Haus (von Tom) gebaut.

The friend gets a house (by Tom) built.

´The friend gets built a house (by Tom).´

This subdivision is controversial, because it is debatable if the construction of sein + participle II und bekommen + participle II can be declared as passive sentences. These constructions have the same morphology as the werden-passive and fulfil the function of an optional reduction of the agent which is why these constructions can be used to present an event without focussing the agent (if the agent is reduced) or with a stronger focus on the agent (if the agent is reconstructed).

The construction bekommen + participle II can be classified as a colloquial version of the werden-passive and the verb bekommen, which can be translated in English as to get or to receive, therefore as comparable with an auxiliary verb (Brinker 1971: 119). Certainly, the German bekommen-passive needs a dative object in the contrasting active sentence (Zifonun et al. 1997: 1825) and the conversion of the object can only be realised by the patients (Eisenberg 2013: 123). Because the construction is very rare in the corpus data and therefore statistically negligible, it is not included in the presented research.

The construction sein + participle II instead is frequent and can be interpreted as the result of a process and thus as a condition, while the construction with werden + participle II expresses a process (Heidolph et al. 1981: 542; Helbig and Buscha 2013: 143; Lenz 2006: 52), as the example in (3) b. shows. Certainly, adjective and verbal participles have the same structure and can be used in the same structure of sentences in German language. This is the reason why a construction can not clearly be categorised as a sein-passive or a copula construction, if the agent is not reconstructed.

Maienborn (2007: 92–97) describes opportunities for suitable analysis and tests: It is a sign for an adjective participle if the negation is formed with –un (Das Haus ist ungebaut; The house is unbuilt) instead of nicht (Das Haus ist nicht gebaut; The house is not built), if the participle can build a compound with nominal or adverbial elements (Das Haus ist selbstgebaut; The house is self-made), if it can have a comparativ (Die Fenster sind geputzter als der Boden; The windows are more cleaned than the floor), if a coordination with adjectives is possible (Der Boden ist hell und gefliest; the floor is light and tiled), but not with perfect-participles (*Der Boden ist aufgearbeitet und gefliest; The floor is reprocessed and tiled). This ambiguity of constructions can lead to the assumption that the sein-passive is less verbal than the werden-passive and less adjective than the copula construction. This observation confirms that “the distinction between the active and the passive is not a strict dichotomy but a gradient ranging from fully verbal to fully adjectical” (Gilquin & Granger 2021: 72).

Considering these aspects, the constructions of sein + participle II can not clearly be classified in the corpus linguistic analysis without subjective interpretations of the meaning of verbs and/or participles. In addition, it can not be excluded that structures with werden + participle II and structures with sein + participle II are processed differently by students because of the different location on the continuum between verbal and adjective. Accordingly, the corpus linguistic analysis and the survey only focus on structures with werden + participle II.

2.2 Corpus linguistic analysis

The following research questions are fundamental for the corpus linguistic analysis, which is presented in the following:

1. How does the relative frequency of passive-sentences differ between contexts and languages?

2. How does the semantic of the agent of passive sentences differ between contexts and languages?

For the investigation, a corpus was generated by Olthoff (2021: 126–230) consisting of texts which are targeted for teenagers in secondary school level. These are three official school books for history and three official school books for biology for secondary level. Furthermore, the corpus includes texts with leisure contexts for teenagers, which are ten editions of the youth magazine Bravo, the book Harry Potter and the Philosopher’s Stone in German translation. The data were complemented afterwards by analysing five extracts of official school books for chemistry (Olthoff and Romstadt 2023). To enable a comparison across languages, this paper involves consideration of a school book for biology in the English language, which is published for the use in bilingual classes in Germany.

For the corpus linguistic analysis, phraseological units are classified as a fundamental item for further considerations (Olthoff 2021: 145). Thus, all phraseological units are initially categorised in two clear categories by answering the question, if it is a passive or not. Chapter 2.1 shows that passive voice in English and German uses different auxiliary verbs which leads to a varying level of demarcation. While passive sentences in English language can be distinguished clearly from other sentences, the delimitation in German language is vague and gradual. Therefore, the analysis of German texts includes only werden-passive sentences in the corpus which can be categorised as prototypical passive sentences. The analysis of the school book in English language includes all constructions of a conjugated form of be and participle II as sentences in passive voice. This might increase the relative frequency of passive in the English school book, because the analysis includes sentences which are excluded in the German texts as potential copula constructions (see examples in 2.1).

Afterwards, the phraseological units, which are categorised as a passive, are analysed regarding the semantic of the (reduced or restructured) agent by using a developed grid-system from Olthoff (2021: 109–120). This system is oriented towards the scale of animation by Silverstein (1976), which is discussed by Kittilä et al. (2011: 5) as well. The following semantic categories are determined: human, personification, animal, plant, object/process, system, personification, unknown (Olthoff 2021: 121). The exemplary results here focus on the most relevant and most recurrent categories ´human´ and ´system´, which is why they are illustrated by examples from school books in the following:

(4) a. human agent in schoolbook for history (Bakenhus et al. 2015: 28):

Die Anführer [...] wurden hingerichtet.

The leader were executed

´The leader were executed.´

b. systemic agent in schoolbook for biology (Bergau et al. 2015: 115):

Wasser wird in Moospolstern gespeichert.

Water is in moss cushions stored

´Water is stored in moss cushions.

c. systemic agent in schoolbook for chemistry (Olthoff & Romstadt in print: 14):

Moleküle werden durch Wasserstoffbrücken zusammengehalten.

molecules are by hydrogen bonds connected

´molecules are connected by hydrogen bonds.´

In the first example in (4) a., which is from a German school book for history, the human agent is reduced, therefore not named, but still able to act. That shows that the human agent has the ability to take action, regardless of the structure of arguments (Dürscheid 1997: 245). In this case, it is not important who conducted the process of execution. The focus is on the process itself. The relevant information for the reader is that the leaders paid with their lives for something. Of course, the process is still a targeted and conscious activity of a human, which might be the result of the semantics of the verb as well. The reconstruction of an inanimate object (f.e. *Die Anführer wurden vom Messer hingerichtet; The leaders were executed by the knife) or an inanimate process (f.e. *Die Anführer wurden von der Gerechtigkeit hingerichtet; the leaders were executed by the justice) are strange and hard to imagine. This confirms that even if the agent is not named explicitely and therefore reduced it is still existing but presented implicitly. For a correct reception of the meaning of the sentence in (4) a. it is not important to know who exactly executes but it is important to know that there is a human ´who´ who conducts the action.

The sentences in (4) b. and (4) c. differ from the sentence in (4) a: Do these sentences include an agent at all? Is somebody or something even acting here? Is even a process described? The circumstances in (4) b. and (4) c. happen systematically and are understandable if they are interpreted as a state condition and not as a process. The reconstruction of a human (*Das Wasser wird von Peter in Moospolstern gespeichert; The water is stored in moss cushions by peter) or an animate agent is not possible, because the storage of water in moss cushions or the coherence of molecules can not be a conscious or targeted activity from an acting individuum. Leong Ping Alvin (2014: 10) lists comparable sentence from a corpus of english scientific research articles: “... a strict monolayer of carbon atoms tightly packed into a hexagonal honeycomb lattice, and fullerene, ...” and points out, that there does not exist an agent which can be indicated, wherefore a reformulation of sentence in active voice “results in some awkwardness”. Certainly, the structure of the passive allows us to not mention the agent while an agent in active voice needs to be named because it is linked to the mandatory subject. This could be a reason why the structure of passive voice is utilised to express those abstract states without an agent.

The exemplary sentences from the analysed school book for biology in English language in (5) show that the frequent semantic categories which are described above are representative at this source as well.

(5) a. human agent in school book for biology in English language

(Bächle-Knauer & Bächle 2005: 11):

Counting chambers are normally used in medical laboratories to determine the number of blood cells in a sample.

b. systemic agent in school book for biology in English language

(Bächle-Knauer & Bächle 2005: 15):

The cells in the upper section of the human skin (epithelium) are arranged in multilayers.

The sentence in (5) a. show that human agents are used in school books for biology in the English language. Specific in this example is that a preposition group gives additional information about the location of the described action. By knowing that the relevant places are medical laboratories, the reader can reconstruct from their global knowledge that the acting people are those who work in these laboratories.

Instead, the sentence in (5) b. is analogous to the sentences in (4) b. and (4) c.: The arrangement of cells can not be conducted by concrete and animated creatures. The described scenario is not even a performed activity; it is rather a static condition on a microscopical level of viewing which is kind of abstract. Thus, abstract issues without concrete agents can be expressed linguistically by passive sentences in German and English language, because this structure offers the opportunity to skip the agent.

The data for the relative frequency of werden-passive sentences and the distribution of human and systemic agents in German school books for the school-subjects biology and history as well as those data for texts with a leisure context are taken from Olthoff (2021: 145–213); the data for chemistry school books are from Olthoff and Romstadt (in print). These results are complemented by an analysis of 20 pages of the biology school book for bilingual classes in English from Bächle-Knauer and Bächle (2005).

The data is evaluated by using the units of central tendency mean value and standard deviation. Correlations to the context are recorded by applying Pearson’s chi-square-tests (χ²-tests).

2.3 Survey of students

The results of the corpus linguistic analysis can be used for a description of the factual use of passive sentences in written texts from different contexts. This allows no conclusion for the understanding of passive sentences of students. Unfortunately, it is difficult to explore, if the understanding depends on one concrete syntactic structure. Every sentence includes specific vocabulary, has to be presented in context for an authentic selection and survey methods work mostly with language – to point out just a few variables which can influence the results of a comprehension test. To approach the issue of language-based challenges in written texts for students, the students were asked by Olthoff (2021: 219–226) regarding their impression of the profile of requirements. Regarding this, the following questions are the basis for a survey of students:

1. Are werden-passive sentences and active sentences classified differently by students in German schools regarding their level of difficulty?

2. Are werden-passive sentences with a human agent and werden-passive sentences with a systemic agent classified differently by students in German schools regarding their level of difficulty?

To answer this question, a survey is distributed in two German classes in grade 8, which builds on the results of the corpus analysis.

The questionnaire presents different sentences (in total 16 sentences) to the students and asks them to rate the degree of difficulty of these sentences for younger children on a four-stage scale. The sentences are designed regarding the results of the corpus analysis, so that concrete answers are possible to the assessment of separate characteristics. Other aspects as vocabulary, length of words and sentences, syntactic structure or the reduction or reconstruction of the preposition group were controlled. Furthermore, the presented sentences are arranged differently and consciously in four designs of questionnaire. In total, data of 23 students are available and included in to the analysis. The sample is certainly too small for general interpretations and does not raise this claim. As a pilot study, the raised data can show first tendencies and can be used as a basis for further studies. Calculations of correlations to characteristics are conducted by a friedmann-test (Olthoff 2021: 213–230).

3.1 Frequency of passive sentences

In the first instance, the results of the quantitative distribution of the passive sentences are presented. Table 1 presents the absolute number of phraseological units and the absolute frequeny, the mean value and the standard derivation of phraseological units in passive voice in each examination unit and Figure 3 illustrates the data of central tendency in a boxsplot.

The data shows that the relative frequency of the passive differs between contexts in German language: The texts from everyday context differ significantly to school books for history (χ² = 1294.45; df = 1; p < 0.001; critical χ²-value = 10.83; w = 0,16), to school books for biology (χ² = 1501.10; df = 1; p < 0.001; critical χ²-value = 10.83; w = 0.17) and to school books for chemistry (χ² = 2241.32; df = 1; p < 0.001; critical χ²-value = 10.83; w = 0.24). Additionally, the school books for chemistry show a higher proportion of the werden-passive compared to the other school books in German language, which is significant for history (χ2 = 168.64; df = 1; p < 0.001; critical χ2-value = 10.83, w = 0.09) and biology as well (χ2 = 154.98; df = 1; p < 0.001; critical χ2-value = 10.83, w = 0.08). Noticable but not surprising is that the detected value of of the biology school book for bilingual classes in English language is above the values of German texts and has a percentage portion of 23.65 %, which is a significant difference (χ2 = 40.82; df = 1; p < 0.001; critical χ2-value = 10.83, w = 0.08). As explained

Figure 3

Relative frequency of phraseological units in German werden-passive and English passive in different contexts in a boxplot.

in chapter 2.1, passive voice is built with different auxiliary verbs in Englisch and German language. While the clarification of passive sentences can be made easiliy and clearly in English language, the use of different auxiliary verbs in German language, which is accompanied by a gradual and in some cases ambiguous classification, leads to the nesessity of methodological focus on werden-passive. Sentences built with sein or bekommen and participle II are excluded but may be convergent with included passive sentences in the school book for biology in English language. In summary, the linguistic differences between Englisch and German lead to potencial anomalies in the data.

Regarding the potential challenge of passive sentences, Olthoff (2021: 237–258) shows that the assessments of sentences in passive or active sentences by students differ not significantly (U = -0.061; p = 0.951; N_Students = 23). This leads to the assumption that sentences are not classified as demanding by students because of the general use of passive. In retrospect, knowledge about the absolute number and the relative frequency of passive sentences is not the final solution for further interpretations regarding potencial language-base challenges of students. Further caracteristics of passive sentences are needed. A relevant aspect can be the semantic of the agent in passive sentences which is considered in chapter 3.2.

3.2 Semantic of agents in passive sentences

The results of the absolute numbers and the distribution of human and systemic agents in passive sentences are presented in table 2 and in figure 4 and figure 5 in the form of boxsplots.

Figure 4

Percentage frequency of phraseological units in werden-passive with human agents in different contexts.

A human agent is in all contexts the largest proportion, which indicates that human agents are prototypical in passive sentences. Certainly, the quantity of passive sentences with a human agent varies depending on context: The amount is lower in school books for natural science, whereat the mean value is located between 40.47 % for biology school books and 68.62 % for chemistry school books, than in school books for history, with a mean value of 90.10 % or texts for leisure contexts with a mean value of 80.82 %. The frequency in the biology school book in English language is 51,81 % and therefore between school books for biology and chemistry in German language.

Figure 5

Percentage frequency of phraseological units in werden-passive with systemic agents in different contexts.

These differences are connected automatically with the proportion of systemic agents, which can be indicated as the second biggest category of semantic agents in passive sentences (Olthoff 2021: 194). Systemic agents are not existing in the analysed school books for history and with a mean value of 0.75 % very low in texts with a leisure context. In the school books for natural science in German language, semantic agents in passive sentences are an extensive realised category, which is located between a mean value of 24.57 % in chemistry school books and 32.65 % in biology school books. Regarding this, the proportion of werden-passives with a systemic agent in school books for biology differs significantly to the proportion in school books for history (χ2 = 493.26; df = 1; p < 0.001; critical χ²-value = 10.83; w = 0.42), significantly to text with a leisure context (χ2 = 208.44; df = 1; p < 0.001; critical χ²-value = 10.83; w = 0.31) but not significantly to school books for chemistry (χ2 = 5.88; df = 1; p = 0.015; critical χ²-value = 10.83). This leads to the suggestion that systemic agents are characteristic in scientific texts in German language. Certainly, it is noticeable that the analysed biology school book in English language shows a salient higher proportion of systemic agents in passive sentences with a mean value of 48.19 %, which leads to a χ2- value below the critical χ²-value (χ2 = 0.63; df = 1; p < 0.001; critical χ²-value = 10.83), but is converted to a non significant p-value of 0.55 because of potential inaccuracies of measurement. Therefore, systemic agents in passive sentences are not specific for scientific texts in German language. This additional sub-characteristic can be indicated as a phenomenon across languages.

Moreover, the assessments of students in the survey differ significantly between passive sentences with a human agent and passive sentences with a systemic agent (U = -3.27; p = 0.001; N_students = 23). Consequently, the classification of the semantic of the (implicit and explicit) agent in passive sentences is a relevant information to specify what exactly is a potential challenge of passive sentences for students.