An important part of academic argumentation is not what you say, but how you say it. It’s one thing to make a bold claim, and another to “soften” it by adding expressions like or something like that, more or less, or in a way. These recurring chunks aren’t merely filler – they convey important interactive information. Vague expressions, or VEs, “express the speaker’s uncertainty or personal attitude towards the proposition and indicate for example solidarity” (Metsä-Ketelä 2012: 264).
Earlier research has expressed concern about non-native speakers’ learning and use of vague expressions, with the danger of sounding “blunt” or “pedantic” if these VEs are underused. In a recent paper by ELFA project member Maria Metsä-Ketelä, these concerns were investigated in the ELFA corpus of spoken academic ELF (English as a lingua franca). How are these vague chunks employed by second-language users in interaction with each other, and how do these findings compare to similar native-speaker data?
OI chunks: organising interaction
You’ll notice that vague expressions like and so on and in a sense function as units – they’re fixed chunks of language that typically don’t vary in form. From a Linear Unit Grammar (LUG) point of view, these are OI chunks (Organising Interaction) which can be used by a speaker to qualify her stance on the main content of an utterance. As Maria points out, the vague expressions in her study serve to intentionally add imprecision. They also have two other important traits:
- VEs do not contribute to the propositional content of an utterance, or the message itself (the M chunks in LUG)
- VEs “are supplementary, that is, they could be omitted from the utterance without compromising its syntactic structure” (Metsä-Ketelä 2012: 265).
These are good descriptions generally for the OI chunks in LUG. You can refer to my earlier introduction to LUG for more background, but this is how these vague OI chunks appear in LUG notation (hover the mouse over chunk labels to see their full description):
MA they grow in
+M-in a world
+M-they constantly mediate
+M with adult knowledge
+M adult ways of living
OI and so on
In this example from the ELFA corpus, the Message chunk with adult knowledge forms the first point of completion, with an additional completion point of adult ways of living. The vague expression and so on comes at the end of the sequence as a supplementary chunk, also providing a moment for utterance planning and further signalling to listeners that this is a possible place for speaker change.
These vague OI chunks aren’t, however, confined to the “end” of a complete sequence of Message chunks. They can appear “between” the Message chunks, as in this example from ELFA:
OImore or less
+Min the same way
Finally, you can also find vague OI chunks at the start of a new chain of Message chunks, like this speaker describing the status of teachers in Finland:
OIin a way
MSin the professionalism
MSof the teachers
+Mis still big
MSin this country
Frequencies of vague expressions in academic ELF
Maria compiled a list of these vague chunks from earlier studies on vague expressions, and searched for them within a subcorpus of ELFA totalling 613,672 words (~80 hours of speech). The speech events are mainly dialogic (two-thirds of transcribed words), with one-third of the data drawn from monologues such as lectures and presentations. These unelicited, naturally occurring events were recorded in two universities in Tampere, Finland.
The vague expressions (VEs) turned out to be frequent, with an overall standardised frequency of 27 occurrences per 10,000 words of data. Maria also found a good deal of interspeaker variation. For example, the expression let’s say occurs five times per 10,000 words in her data, but these are concentrated in just three speech events. Similarly, in a way occurs 173 times in her data, but a third of these instances comes from a single speaker in a doctoral defense.
Insofar as these VEs are interactive chunks (OI), it’s no surprise that Maria found a statistically significant higher frequency of VEs in dialogic speech events (30 VEs per 10,000 words) vs. monologues (22 VEs per 10,000 words). So far, nothing too shocking jumps out from these numbers. But what about the native speaker comparison data? Shouldn’t ELF speakers “underuse” these chunks in relation to English native speakers?
Damned if you do, damned if you don’t
For comparison data, Maria searched for the same vague expressions in the MICASE corpus (Michigan Corpus of Academic Spoken English). The table at right shows her findings for the 10 most frequent VEs in each corpus, sorted by frequency per 10,000 words (see Metsä-Ketelä 2012: 277). Comparing these top-10 lists, seven of the VEs are identical, though with different rank orders and frequencies; the overlapping expressions are highlighted in yellow. So the picture is quite similar, though culture-specific preferences emerge – and stuff (like that) is fifth on the MICASE list, but is only found 15 times in Maria’s ELFA subcorpus.
Turning to the overall frequency of these top-10 vague chunks, they’re used almost twice as often in the ELF data (19 per 10,000 words) than in the North American data (10 per 10,000 words). Surprising? This distribution holds up when all vague expressions in the study are taken together (excluding outliers of less than five instances), with 25 VEs per 10,000 words in ELFA and only 14 VEs per 10,000 words in MICASE. Clearly, US English speakers dramatically underuse these vague chunks and they must be helped.
This sounds funny, since for many linguists, imitation of “the” English native speaker is the true goal of using English. Learner language research is based on the assumption that non-native speakers inevitably “underuse” or “overuse” some linguistic feature in relation to native speaker corpora. By design, there is always something “wrong” that pedagogues must fuss about, since that’s what keeps English teachers and testers in business – a product that’s always broken.
But if you let ELF data speak for itself, it’s clear that ELF users are aware of these vague organising chunks, the functions they serve in interaction, and how and where they are deployed. They’re used similarly to native speakers, but more frequently. Why? I would argue that all speakers have some stock of pre-fabricated, formulaic chunks in memory. Naturally, a second-language speaker’s stock will be less extensive than a native speaker’s. The most likely chunks to be stored by ELF users are the highest frequency, formally fixed chunks – the ones we hear repeatedly and that don’t tend to vary in form. These are precisely the type of organising chunks under discussion.
So perhaps it’s no surprise that academic ELF speakers use these vague organising chunks more often than English native speakers. It’s up to you to decide if an evaluative judgment should be attached to this observation. But for a take-home message, I’ll let Maria have the final word:
… at least when it comes to academic discourse, there is no need for concern about non-native speakers sounding ‘blunt,’ ‘pedantic,’ or ‘bookish’ due to their inability to include appropriate vague expressions in their utterances, as has been suggested in previous studies.
(Metsä-Ketelä 2012: 282)