7.1 Stereotypes about grammatical complexity: What does it mean to be ‘complex’?

Traditionally, linguists have operationalized grammatical complexity as the increased use of ‘elaborated’ structures, consisting of dependent clauses added on to a simple independent clause. That is, a sentence is considered complex to the extent that it has multiple embedded subordinate clauses.

We do not dispute the claim that ‘elaborated’ clauses with embedded dependent clauses represent one type of grammatical complexity. However, the central themes in our book include the complementary claims that:

• – there are different types of grammatical complexity;

• – one of these alternative types of complexity involves the extensive use of embedded phrasal structures (rather than embedded clauses);

• – phrasal complexity features are much more characteristic of, and thus more important in, academic writing than clausal complexity features; and

• – these phrasal complexity features have been the locus of historical change in academic registers over the last three centuries.

We have amply illustrated how phrasal embedding – especially phrasal features functioning as noun phrase modifiers – can result in sentences that are grammatically complex. We have further shown that these kinds of complexity are less explicit, and therefore more complex from a processing perspective, than alternative structures with dependent clauses. And we have shown that these phrasal complexity features are highly specialized: they are relatively recent innovations, and they occur primarily in academic (science) research writing. That is, these phrasal complexity features are ubiquitous in that register, resulting in discourse passages like the following:

Each of the sentences in this passage has a single main verb (is often looked, is, applies), and only one of these sentences has a dependent clause (with the non-finite verb studied). If we restricted our definition of complexity to consider only the use of dependent clauses, we would conclude that this passage was not complex – a conclusion completely at odds with our perceptions as readers! This apparent contradiction reflects the fact that the sentences in this passage are in fact very complex in their reliance on phrasal modifiers of noun phrases, including:

• – pre-modifying nouns (e.g., cell membrane, electron microscope investigations),

• – attributive adjectives (e.g., preparative treatment),

• – prepositional phrases as post-nominal modifiers (e.g., a basis for the interpretation…, a difficulty in these investigations…),

• – appositive noun phrases (e.g., the strongly oxidizing fixatives: osmium tetroxide and potassium permanganate).

One of our main goals in the book has been to advocate this alternative perspective on grammatical complexity, showing that embedded phrasal structures are at least as important as embedded dependent clauses for characterizing grammatical complexity. In addition, we have shown that grammatical complexity is not a single unified construct. Rather, there are multiple types of grammatical complexity, and these types differ in important ways with respect to their structural and syntactic characteristics, their discourse functions, and their distribution across registers. Accordingly, any attempt to analyze grammatical complexity through a single omnibus measure is misguided. Quantitative measures such as sentence length or t-unit length – motivated by the belief that there is a unified construct of grammatical complexity – fail to capture the important differences associated with clausal versus phrasal complexity features. This recognition is important both theoretically and practically. We return to the applied implications of these research findings in Section 7.4.

7.2 Stereotypes about historical change: Does all grammatical change originate in speech?

In Chapter 1, we documented how linguists have often staked out an extreme position on the status of written discourse: speech is often described as ‘real language’, while written discourse is often dismissed as a secondary artifact. When it comes to the study of historical change, linguists complain that we are stuck with written texts as our only indication of language use in earlier centuries. This is often portrayed as an unfortunate situation, because direct analysis of written discourse is claimed to have little theoretical interest in itself. Rather, the assumed task of the historical linguist is to remove the obscuring filter of the written form, to discover historical change as it really occurred in speech, which is assumed to be the real language underlying written discourse. Some linguists acknowledge that written discourse also changes, but the generally accepted view is that speech takes the lead in linguistic innovations, with written discourse trailing along behind: ‘the spoken form always, always keeps on changing–and ultimately drags the written form reluctantly with it’ (McWhorter Reference McWhorter2001: 17). Academic writing is regarded as especially conservative and resistant to historical change: one of the ‘up-tight’ registers identified by Hundt and Mair (Reference Hundt and Mair1999).

In the present book, we have challenged these stereotypes at every level. We adopt the ‘register’ perspective on language use (see Biber and Conrad Reference Biber and Conrad2009), which begins with an analysis of the situational context of a text, followed by a linguistic analysis of the distinctive grammatical characteristics of the text. The patterns of linguistic variation among texts are then interpreted in functional terms, based on the claim that:

From this perspective, all registers (spoken or written) will have distinctive linguistic characteristics, which are functionally associated with their situational context of use.

The register perspective can be considered as an extension of earlier research carried out within the general framework of ‘functional linguistics’:

Research carried out within the framework of ‘functional linguistics’ has usually been restricted to language use in spoken interaction. In contrast, the perspective of register variation adopted in the present study is based on the premise that every register (written as well as spoken) has a situational context and communicative purpose that functionally ‘motivates, constrains, explains, or otherwise determines grammatical structure’ (Nichols Reference Nichols1984: 97). It is only logical that these functional forces operate diachronically as well as synchronically. As a result, far from being a filter that obscures language change, we have argued that written discourse provides the context needed for certain types of linguistic change – that is, that certain types of linguistic change occur in writing due to the situational context of written language. In particular, we have shown how the informational purposes, careful production circumstances, and specialist audiences of scientific research writing are functionally associated with dramatic historical innovations and functional extensions for a range of phrasal complexity features.

7.3 Stereotypes about academic writing: Is it resistant to change, elaborated, and explicit?

It turns out that academic writing is especially distinctive relative to the range of other registers: distinctive in the types of grammatical complexity employed in texts, and distinctive in the patterns of historical change. And so the present book has focused on the grammar of academic writing – and as a result, challenged multiple stereotypes about that register.

For example, linguists have characterized academic writing as especially conservative and resistant to historical change, contrasted with popular written registers that are supposedly more receptive to change. This stereotype persists because linguists have assumed that all historical change originates in speech, and thus they have considered only linguistic innovations that originated in speech. With that restricted focus, the stereotype is accurate: the findings reported in Chapter 4 (see Table 4.2) show that colloquial innovations like semi-modals and phrasal verbs are adopted quickly in popular written registers like fiction, while they have generally not been adopted in academic prose. Thus, academic writing is indeed resistant to change, if we consider only grammatical features that originate in speech.

However, when we entertain the possibility that grammatical change can originate in writing, we start to notice completely different kinds of linguistic features, and we discover that academic writing has been anything but conservative and resistant to change. In fact, academic writing has been the leader in these other kinds of historical developments, employing grammatical patterns of use that are only gradually becoming adopted in popular written registers. When the full suite of phrasal complexity features are considered, we see a dramatic historical shift in grammatical discourse style that is unprecedented in any other register: moving away from the heavy reliance on verbs and dependent clauses in the eighteenth century, to the predominance of noun phrases and phrasal modifiers (coupled with the relatively rare use of verbs and dependent clauses) in the late twentieth century.

As noted in Section 7.1, many linguists operate with the stereotype that grammatical complexity is equivalent to elaborated grammar, which can be operationalized as the dense use of dependent clauses. Because academic writing is complex in content and takes a long time to learn, it has simply been assumed that it must be structurally elaborated – and thus, must employ many dependent clauses. We have shown that this stereotype is also fundamentally incorrect: academic writing is indeed complex, but it is structurally compressed rather than elaborated, and the predominant grammatical complexities of academic writing are phrasal rather than clausal. Recognition of these distinctive grammatical characteristics allows us to trace the dramatic historical changes from the elaborated clausal style of eighteenth century academic research articles to the compressed phrasal style of modern research articles.

An associated stereotype is that academic writing is maximally explicit in the expression of meaning. The primary communicative purpose of academic writing is to convey information, but there is normally no direct contact between author and reader, and thus no chance to clarify meanings that are unclear. Given that situation, it seems obvious that academic writers should employ grammatical structures that are maximally explicit in meaning. But they do not. In fact, the compressed phrasal devices employed in academic writing could be described as maximally inexplicit in meaning. These devices omit verbs and grammatical signals that would help readers identify the meaning relations among elements. The resulting compressed structures can be highly efficient for expert readers, but they are certainly not explicit in the expression of meaning.Footnote ¹

All of these peculiar linguistic patterns are especially characteristic of specialist science research writing. Academic writing has often been treated as a single general register, making it easy to disregard the systematic patterns of linguistic variation among academic sub-registers. However, consideration of academic sub-registers is crucially important for recognizing the distinctive linguistic patterns of linguistic variation documented in the preceding chapters. At one extreme, humanities research writing is relatively conservative and resistant to change: to a large extent, modern humanities books and articles employ the same grammatical discourse styles as eighteenth century academic prose, relying heavily on elaborated structures with dependent clauses. In contrast, specialist science research articles have changed radically in their grammatical discourse style, and the innovative phrasal complexity features and the dramatic patterns of historical change documented in preceding chapters are especially characteristic of that academic sub-register.

The discussion in Sections 7.1–7.3 have primarily focused on the theoretical implications of the findings presented in this book, challenging commonly-held perspectives on the nature of academic writing, the study of linguistic change, and the nature of grammatical complexity in English. However, these linguistically- and research-oriented discussions naturally prompt a consideration of the resulting implications for more applied areas of linguistic research. We turn to a brief discussion of these applied concerns in Section 7.4, demonstrating a range of connections between the empirical patterns documented throughout the book and the interests of applied linguistics working in language teaching, learning, and assessment.

7.4 Implications for applied linguists: From language development to language teaching and assessment

The linguistic findings presented here also challenge the stereotypes and practices of applied linguists. For example, studies of language development (especially writing development) have been based on the assumptions that grammatical complexity is a unified construct, that it can be measured through a single omnibus measure (usually based on ‘t-units’), and that the increased use of dependent clauses represents increased proficiency. Practices for language teaching and assessment are based on these same assumptions. In contrast, the compressed discourse styles actually found in advanced academic writing are often disregarded in the teaching and testing of academic reading and writing.

Taking a register variation perspective, coupled with detailed consideration of phrasal complexity structures, leads to fundamentally different expectations about language development. We have discussed the applied implications of our research in several previous publications (e.g., Biber and Gray Reference Biber and Gray2010; Biber, Gray, and Poonpon Reference Biber, Gray and Poonpon2011, Reference Biber, Gray and Poonpon2013; Biber, Gray, and Staples Reference Biber, Gray and Staples2014). For experts in an academic sub-discipline, the compact, inexplicit discourse style of research articles is functional and efficient. By compressing information into few words, specialists can quickly read through a text and obtain the information they need. The lack of explicitness rarely causes problems, because specialists have extensive background knowledge in the subject area. However, the situation is the opposite for novice students. Students lack the specialist knowledge that would allow them to readily infer the expected meaning of compact, inexplicit grammatical constructions. Thus, there are clear implications of these findings for the teaching of academic reading: whether we like these discourse styles or not, it is a certainty that students will encounter professional written texts with these characteristics, and that students’ academic success will depend on their ability to extract the intended meaning from these texts.

Traditional grammar practice for advanced academic purposes tends to focus on elaborated structures, usually dependent clauses like relative clauses, adverbial clauses, and complement clauses. Phrasal modification – especially noun phrase structures with multiple levels of phrasal embedding – is given much less attention. This practice might in part reflect a perception that phrasal modification is somehow easier than clausal embedding. However, as documented in the preceding chapters, the opposite is in fact the case: meaning relations are much clearer and more explicitly expressed through clausal embedding than phrasal embedding. The applied implication is that English for academic purposes (EAP) reading courses should provide extensive practice in decoding texts with extensive phrasal modification (see, e.g., Musgrave and Parkinson Reference Parkinson and Musgrave2014 on teaching nouns as nominal pre-modifiers).

The applied implications for academic writing instruction are more complicated. When teaching advanced academic writing, both ‘clarity’ and ‘economy’ are important. There is a fundamental tension between these competing goals. Students need to learn how to write prose that is clear, where the reader has no doubts about the intended meaning. But at the same time, compact prose is valued in advanced academic writing (especially science writing), and a high level of implicitness is tolerated by expert readers. It is not clear how these competing demands are best handled in advanced EAP writing courses, but given that nearly all academic research writing has these characteristics, this is clearly an area that requires additional research.

In Biber, Gray, and Poonpon (Reference Biber, Gray and Poonpon2011: 30–31), we hypothesized a series of developmental stages for students of advanced academic reading and writing, mirroring the progression from the complexity features that are common in conversation to the complexity features that are common in academic writing. Table 7.1 summarizes the grammatical features associated with each stage. As in the present book, the hypothesized stages are based on consideration of both grammatical form and syntactic function. Thus, the stages generally progress from finite dependent clauses functioning as constituents in other clauses, through intermediate stages of non-finite dependent clauses and phrases functioning as constituents in other clauses, and finally to stages requiring dense use of phrasal (non-clausal) dependent structures that function as constituents in noun phrases.

In addition, we considered lexico-grammatical factors in positing the developmental stages, because dependent clause structures in conversation tend to occur with just a few controlling words, while structures in academic writing occur with a much wider range of controlling words. For example, 75% of all that-complement clauses in conversation occur with only four verbs: think, say, know, and guess (see Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 668). To-clauses are generally not frequent in conversation, but the combination want + to-clause is extremely common (see Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 711). We thus hypothesize that these relatively fixed lexico-grammatical combinations are acquired at an earlier stage than full control of the target syntactic structure.

Recent studies of writing development provide support for these hypothesized progressions. For example, Taguchi et al. (Reference Taguchi, Crawford and Wetzel2013) use many of these same features to investigate L2-English student writing, finding that lower-rated essays used more finite/non-finite dependent clauses while higher-rated essays used more attributive adjectives and post-noun-modifying prepositional phrases. Parkinson and Musgrave (Reference Musgrave and Parkinson2014) utilize a subset of the developmental stages from Biber et al. (Reference Biber, Gray and Poonpon2011) to compare L2 writing by EAP students and matriculated MA students. The results provide strong support for certain aspects of the hypothesized developmental progression at advanced stages of writing development. For example, EAP students used many more attributive adjectives than MA students, while MA students used twice as many pre-modifying nouns as EAP students (see Parkinson and Musgrave Reference Musgrave and Parkinson2014: figure 1, p. 56). Similarly, post-nominal prepositional phrases were much more common in MA writing than in EAP writing.

Biber, Gray, and Staples (Reference Biber, Gray and Staples2014) investigated a large set of clausal and phrasal complexity features within the context of a standardized language exam (TOEFL-iBT). That study failed to find systematic differences among score levels for the use of individual complexity features. However, the first parameter of variation identified in a multi-dimensional analysis provided strong support for the basic opposition between phrasal versus clausal styles of complexity: the ‘positive’ features on that dimension included nouns, attributive adjectives, pre-modifying nouns, and prepositional phrases; the opposing ‘negative’ features on that dimension included verbs, that-complement clauses, and finite adverbial clauses (see Biber, Gray, and Staples Reference Biber, Gray and Staples2014: table 5).The differences among score levels (and task types) were highly systematic with respect to this linguistic dimension, with higher proficiency test-takers employing more nominal/phrasal discourse styles than lower proficiency test-takers.

The results of these studies suggest that the entire suite of complexity features work in concert. Thus, while it might be difficult to isolate progression in the use of an individual feature, it is more likely that there will be important differences in the use of the full set of co-occurring complexity features. In addition, comparison of the Biber et al. and Parkinson and Musgrave studies suggests that these developments are best observed at very advanced stages of writing development. Thus, there appear to be only minor differences among proficiency levels in beginning EAP courses and undergraduate composition courses, because few students employ phrasal discourse styles at that level. At more advanced levels, though, we begin to observe the predicted developmental progressions in the use of phrasal complexity features.

Obviously, much more research is required to investigate these possibilities. The present book has provided the linguistic foundation for that applied research: challenging the profession to rethink notions of grammatical complexity and incorrect stereotypes about professional academic writing. Future research is required to track the actual paths that students follow as they move from competence in the use of clausal complexity features found in conversation to competence in the use of phrasal complexity features found in advanced academic writing. Building on that applied research, we expect that future practice in language teaching and testing can better account for the actual patterns of language use found in advanced academic writing.

7.5 Conclusion

Our goal in the present book has been to show how the corpus-based analysis of historical linguistic change in academic writing challenges numerous widely-held beliefs and stereotypes. Rather than being restricted to spoken discourse, we have shown how major historical innovations and developments can occur in writing. Rather than being conservative and resistant to change, we have shown how specialist academic research writing has been the locus of some of the most dynamic grammatical changes in the last two centuries. For the most part, those changes involve the development of phrasal complexity features. Thus, rather than focusing on dependent clauses as indicators of grammatical complexity, we have argued that phrasal modifiers are equally important – although fundamentally different – reflections of complexity. As a result, rather than being structurally elaborated, we have shown that the hallmark grammatical style of academic research writing is characterized by structural compression. And finally, rather than being maximally explicit in the expression of meaning, we have shown how academic research writing is often amazingly inexplicit, due to the loss of meaning associated with the use of compressed rather than elaborated structures.

The corpus-based research findings in this book provide strong support for these alternative perspectives. It is our hope that the volume will serve as a starting point: a foundation for future research by scholars coming from a range of perspectives, including theoretical and descriptive linguists, discourse analysts and applied linguists, sociolinguists and historical linguists.