2.3 Dimensions of English

Originally, MD analysis was developed by Douglas Biber to investigate spoken and written English in a range of texts and situations. In the 1988 study, he analysed the English language by means of the LOB Corpus, which comprises about 500 texts from 15 different genres, as well as the London-Lund Corpus of Spoken English, which is a collection of 87 spoken texts. Additionally, he used personal and professional letters not covered by the corpora(cf. Biber 1988: 66). The original MD study of 1988 results in seven dimensions of the English language, which can be applied to new areas of register variation or to study special sub-registers.

In the present section, the results of the original study will be presented in an interactiveonal way. The outcomes of the 1988 factor analysis will be used to interpret the dimensions of English. Table 1 shows the most important features corresponding to the factors as well as the respective factor loadings (features in brackets have a larger loading on another factor).

Table 1: Factors 1 – 7 and factor loadings (Biber 1988: 89-91)

As noted above, the first step for factor interpretation is the assessment of the functions shared by the co-occurring features. Having a look at factor one, the features indicate a quite powerful factor, as it includes numerous features with important loadings and thus represents a very basic dimension of variation in English (cf. Biber 1988: 104).

The interpretation of the negative features seems straightforward. Characteristics like the high frequency of nouns indicate an informational focus, because nouns typically serve the integration of information in a text. The type/token ratio signifies that the texts contain varied vocabulary and words with specified meanings (cf. Biber 1988: 104). That is, apart from the high informational density, the negative features are associated with texts providing a precise lexical choice, such as academic articles (cf. Conrad and Biber 2001: 24).

The positive features are however associated with involved, non-informational conversations, having a more interactive and affective purpose. Features like first- and second-person pronouns, wh-questions, emphatics or amplifiers reflect interpersonal interaction as well as the expression of personal feelings and concerns (cf. Conrad and Biber 2001: 24). The pronouns directly refer to the addressee, as well as the wh-questions which request a direct reaction. The private verbs and present tense forms indicate a verbal style. They refer to actions and expression of private attitudes (cf. Biber 1988: 105). Other positive features signify a reduced structure of language and thus generalised presentation of information, e.g. that-deletion, contractions, pro-verb do or pronominal forms. These lead to a more global and less explicit content (cf. Conrad and Biber 2001: 25).

Thus, factor 1 represents interactional, affective and generalised content versus high informational density and precise content and is therefore named Involved vs Informational Production (cf. Conrad and Biber 2001: 25). The communicative situations influencing this dimension are the speaker’s purpose (informational or involved) and the production circumstances (real-time or edited) (cf. Conrad and Biber 2001: 25). These two characteristics are highly related, that is, real-time discourse with certain constraints usually has an involved, non-informational purpose and vice versa (cf. Biber 1988: 107).

The second step of interpretation is to compute dimension scores in order to distribute different registers along the dimensions. The distribution of the registers of Dimension 1 (Figure 1) confirms the interpretation of involved vs. informational production. The conversational texts are interactive and therefore less informational, because usually the participants do not have informational purposes (cf. Conrad and Biber 2001: 28).

Figure 1: Dimension 1 (Conrad and Biber 2001: 27)

Some registers do not have a clear preference for the features and score therefore on intermediate position. Public conversations or interviews typically have informational purpose but the production circumstance is interactive, as it is a real-time production. On the other hand, academic prose is highly informational, using numerous specified nouns. Additionally, this register is non-interactive and produced under highly controlled circumstances (cf. Conrad and Biber 2001: 28).

The second factor includes a much smaller number of features with important loadings but still represents a rich basis for interpretation. Though the negative features have a stronger influence on other factors, the positive ones characterise the factor sufficiently.

The positive features of factor 2 are all associated with past time narration, which is mostly represented by all kinds of fiction. Especially past tense and perfect aspect forms serve the description of past events (it was difficult to tell, they had given), while third-person pronouns refer to the participants (he, they). Public verbs are used to mark direct and indirect speech (tell, say, explain), which is a common characteristic of fiction. Present participial clauses are used to add descriptions and imagery (looking ahead). As noted above, there are no strongly loaded negative features. Therefore, the factor represents a continuum of the use of positive features (cf. Conrad and Biber 2001: 29). The complementary distribution of the present tense in this factor is due to the fact that narrative discourse tends to deal mostly with past events, while other types of discourse in general have a higher share of present tense references. The attributive adjectives probably reflect the use of elaborated nominal referents (cf. Biber 1988: 109).

Figure 2: Dimension 2 (Conrad and Biber 2001: 32)

The distribution of registers along the dimension supports the interpretation: all types of fiction have high positive scores. Registers concerned with present events or argumentations, such as official documents or broadcasts, have negative scores. Other registers like face-to-face conversations, personal letters or speeches tend to use both narrative and non-narrative discourse. That is why they are clustered around the score 0 (cf. Conrad and Biber 2001: 29).

Figure 3: Dimension 3 (Conrad and Biber 2001: 34)

Due to factor analysis, the subsequent factors become much smaller. Therefore, the interpretation is even more complicated. Factor 3 is dominated by wh- relative clause constructions on object and subject position and pied-piping constructions which identify referents or provide information about referents. The additional positive features suggest informational density and integration (cf. Conrad and Biber 2001: 33). The negative features are however used to refer to places and times outside of the speech context (cf. Conrad and Biber 2001: 33). Therefore, Dimension 3 (Figure 3, above) represents a continuum of texts that have explicit reference and reference that is more dependent on the situational context (cf. Conrad and Biber 2001: 33).

Biber explains that "genres such as official documents, professional letters, and academic prose require highly explicit, text-internal reference, while genres such as broadcasts and conversation permit extensive reference to the physical and temporal situation of discourse." (1988: 143-144)

The fourth factor contains only features with positive weights. Again, it is difficult to interpret a factor with as few features as factor 4. But still, the features represent special functions which are presented by Biber and Conrad: Prediction modals are for example used to show what will be possible in the future or discuss hypothetical situations, whereas necessity modals express obligation for the addressee. Possibility modals further state how likely events are and conditional subordinations describe certain limits or conditions for the situations. The infinitives are often used as adjective complements with adjectives that reflect the author’s stance, while suasive verbs are used to encourage certain actions. All these features thus work to structure an argument. Consequently, the dimension identifies the degree to which persuasion or argumentation is marked overtly. Registers with negative scores thus lack these features and are therefore less persuasive or argumentative (cf. Conrad and Biber 2001: 35; Biber 1988: 111).

Figure 4: Dimension 4 (Conrad and Biber 2001: 36)

Factor 5 is characterised by positive features such as passives. Passive constructions are normally used to express information with little or no emphasis on the agent. Consequently, inanimate referents, which are often modified with past participial passive modifiers, are more important for the discourse (cf. Conrad and Biber 2001: 37). Considering the distribution of registers along Dimension 5, it is obvious that official documents and academic prose are mostly marked by these features. These registers are associated with abstract or technical information. Therefore, this dimension shows the difference between registers with an abstract and non-abstract style (cf. Conrad and Biber 2001: 37).

Figure 5: Dimension 5 (Conrad and Biber 2001: 38)

The last two factors are less important and difficult to interpret due to the few features with important loadings and thus have often been excluded from studies (cf. Conrad and Biber 2001: 39). Factor 6 contains three types of dependent clauses: that-complement clauses on verbs, that-complement clauses on adjectives, that-relative clauses on object position. These are features of structural elaboration which co-occur in this factor with informational features like demonstrative pronouns or stranded prepositions. Conrad therefore argues, that registers with positive scores are informational spoken registers - spontaneous as well as prepared speeches – which make use of these clauses to express personal attitudes (cf. Conrad and Biber 2001: 39). Factor 7 is called “academic hedge” since the few features have functions related to hedging (seem/appear). They are typically used to reflect perception rather than facts. Furthermore, downtoners lessen the degree of certainty. These hedges are often used in academic texts, whereas conversation uses other kinds of hedges (cf. Conrad and Biber 2001: 41).

Created with the Personal Edition of HelpNDoc: Full-featured Help generator