The latest guest blog comes from David Chew, a teacher in the East Midlands. Here he looks at how he approaches the NEA language investigation from its earliest ideas and inception through to the detailed analysis needed to make it work, and he looks at how the mysterious F score can add a new dimension to discussions of formality in texts.
“It’s more maths and science than a literature essay.” This is my opening announcement to a class of English Language students as we embark on the AQA Language Investigation NEA (coursework). The subsequent groans could fill a pandemically-induced empty football stadium.
In an attempt to shift students’ perceptions of having opted for an “Arts” based A level, I riff on about creating “a fair test” and identifying “measures of central tendency”. I am fully aware, after teaching variations of this investigative language study for over 20 years now, that I need to shock Mark into realising that he will need to find something to count, count it and then report what the count tells him about language and language users.
Don’t worry. I hear your gasps of “you can’t reduce the niceties of language analysis to bean counting” and “where are your socio-linguistic sensibilities?”. We will come to that; especially since AO3 is the weightiest of the three AOs awarded to this study and trades marks for appreciation of contextual factors and meanings.
But Mark needs to understand this isn’t an essay; it isn’t a commentary; and it certainly isn’t a report. It is however a precursive experience to a university dissertation.
Assumptions challenged, the next pitfall is approach. There are two processes which students must engage in: conducting an investigation and writing about it. These processes exist in a chicken and egg symbiosis. Do I teach one and assume the other will follow? If they don’t know how the writing will be structured, how can they cover all the bases when they launch themselves into investigating? If they haven’t assembled data and identified variables, how can they formulate a hypothesis? I tend towards spinning both plates at once, knowing that different students will develop their understanding of these processes in different ways.
And what to investigate? I advise students to go with what they know and enjoy. After all, they will be engaging with the material for several months: that’s a prison sentence if your teacher has foisted an idea onto you just because your initial reaction was “I don’t know what to investigate”. So, will it be editorials in Horse and Hounds magazines? Perhaps you’ve noticed that sports commentaries on radio stations are more effusive than TV commentators. Your swimming coach has a different way of addressing the team competitors whether you are winning or losing. You’ve noticed that your young female cousin is learning to read faster than your little brother. You suspect that the talk on reality TV shows featuring young people doesn’t match what you have been told about 20th century theories of genderlect.
However, I draw the lines at poetry and advertising slogans. Not because there isn’t anything to be discovered in these texts: there most certainly is. But you’ve got to write 2000 words covering at least two language levels (or systematic frameworks in old money) and “Guinness is good for you” repeated over the decades with different images of Toucans can only get you so far. Similarly, there is a post grad thesis to be had looking at the implicature of e e cummings dispensing with capitals, but not a successful A level NEA.
There’s always one, though. However much you encourage them to tell you about their latest loot box disappointment, their bilingual grandmother, or their moonlighting gig shelf-stacking on Fridays when they should be attending PSHE lessons, they will still succumb to the lure of an investigation into the comparison of tabloid and broadsheet newspapers. They don’t see any downside to this choice, even when they admit that they don’t read newspapers, and can you just remind them why the Guardian is a broadsheet anyway.
What have I learned over twenty years? A wise colleague transformed my teaching, and students’ outcomes, when she pointed out that moving from a general hypothesis to a detailed analysis was a bit of a stretch for the investigator and the reader. So, the Queen’s Christmas broadcasts have become more informal during her 68-year reign, but what exactly will you look for to support this hunch? This is where a series of language level-based expectations come in. In terms of lexis, there will be fewer Latinate words now than there were in 1952. There will be more colloquialisms in 2020 than ever before. And that use of first-person pronouns, unique to royalty (and Margaret Thatcher), might also have changed semantically. These organised, structured, and coded expectations then become the organising framework and structure for the analysis section. Everything is now set up clearly for the investigator to investigate and the reader to read.
Ideally you would start the investigation at the end of the two-year course. This would allow students to reference theories and theorists which they have already studied in Language and Gender, Child Language Acquisition, power, change, diversity etc. But this isn’t practical, so you’re faced with signposting students to ideas and concepts which they don’t yet know are relevant to their investigation. At this stage you are grateful that, although you don’t know much about anything, you do know a little about everything. There is one theory, however, I discovered that you can bank on to bolster most investigations. One panacea theory; one magic bullet. That is the F Score.
It’s great. It assigns numerical value to word classes based on whether the word class is deemed to be more formal or more informal. So, adjectives are more formal whilst adverbs are more informal. Students look at their data samples, identify the word classes being used, and apply a formula: [F = (noun frequency + adjective freq. + preposition freq. + article freq. – pronoun freq. – verb freq. – adverb freq. – interjection freq. + 100)/2]. Now they have a number for the degree of formality of each data set. We are talking charts, graphs, means, modes, medians, trendlines….. At this stage Mark wishes he had paid attention in GCSE maths. His classmate Sophie did pay attention though and, having analysed a sample of 6 editions of The Aberdeen Press and Journal over 220 years, she has a wealth of statistical analysis about the formality of language.
Even Mark can now see how he could measure the spoken formality of his favourite sports stars:
At this stage you throw in the curve ball. “Well done, Mark. Now which of your subjects are monolingual?” Now you’re sold on this universal remedy, I would love to claim ©dchew, but I can’t. Instead get the full monty here.
Don’t misunderstand me. I’m not an English teacher simply because the Maths department wouldn’t have me. I don’t think that there is beauty in numbers alone; but they do have their place when your student’s investigation needs some close comparative analysis and some marks for AO1 and AO2.
Context really is everything, though. Once you have counted those Latinate derivations, those run of the mill colloquialisms, and “did she mean ‘we’ as a singular or a plural”, you need the context. Why was the 1992 Christmas message an anomaly in the bar chart representation of the Queen’s increasing informalisation? Perhaps because saying “annus horribilis” ad infinitum takes the Latinate lexis count sky high for that year! When those reality TV shows go out after the watershed and editorialise the hours of recorded talk to 45 minutes of the most dramatic dialogues between two self-serving egotists, then perhaps you really do need to consider the Observer’s Paradox. When you have squeezed the pips out of the numbers, you need to recognise that the F score won’t tell you about semantics, production, reception, representation and variables.
Ah, variables. To keep them open or closed? If Mark is comparing female and male language use (go with me on binary for now), he needs data sets from each group. But if he is going to attempt to ascribe any causal links to his findings, then all the other factors in the data set such as age and audience and familiarity and function had better be the same. Every year I find myself trying to explain the implications of correlation and causality to students who would have preferred media studies on their GCSE timetable to pipettes and Van der Graaf generators. I have honed it down to this example which I tell students every year. “I have looked at the latest test results for this class and I have to say I am pleased that on the whole the boys did better than the girls in the class. So, girls, since all the boys wear ties and none of you wear ties, I suggest that you start wearing a tie if you want to be as good as the boys in the next test.”
Four months into the investigation you overhear Mark telling his classmate that trendlines only really work for comparing data sets over time. He then points out that using mode rather than mean would allow the anomalous data set to be included without skewing the results. He berates his friend for ignoring audience demographics and begins to explain synthetic personalisation. You sit back and smile. Your work here is done. The groans which filled that pandemically-induced empty football stadium are no more.