‘ Ow Cockney is Beckham Twenty Years On ? An Investigation into H-dropping and T-glottaling

This research paper examines how language change can occur across the lifespan through the linguistic analysis of East Londoner, and world renowned football player, David Beckham. Specifically, we look at his use of the consonantal variables of t-glottaling and h-dropping and how the frequency of these forms change over a 20-year period. We discuss the background of the linguistic phenomena under investigation and the common environments in which these non-standard variants are likely to occur. We also take a closer look at how the forms are being used in certain phonotactic environments, for example, word-medial and word-final positions, and the potential reasons behind them being less common when preceding or following certain sounds. We discuss some common theories associated with language change across the lifespan, using quantitative data to find trends and qualitative interpretation to suggest social causes for our findings. The paper allows us to critically evaluate language change theories, such as Labov’s (1978) apparent-time theory. In designing our study, we hypothesised that Beckham would be seen to undergo linguistic change from his classic East London Cockney features to more prestigious forms. As t-glottaling and h-dropping are stigmatised forms which are commonly associated with a working-class background, we believed that Beckham would go from using a high rate of these variants in his teenage years, due to his lower socioeconomic background, to producing standard /t/ and /h/ more frequently, reflecting his dramatic upward social climb. Due to his rise to fame, we expected that his celebrity status would bring an added pressure to speak in a “correct” manner, therefore influencing him to opt for the standard variants more frequently. The variants we looked at are also commonly associated with younger speakers, so we expected Beckham’s aging to further affect his language. Our results support our hypothesis, showing the extent to which David Beckham’s language choices have changed over time. We found that he showed a significant decrease in both h-dropping and t-glottaling in all phonotactic environments. However, we also found a surprisingly high rate of t-glottalisation before consonants and after vowels in Beckham’s 2014 recordings. Our data support theories concerning age, social class, sex, and dialect convergence. Overall, our paper offers insights into the methodology and theory surrounding language change across the lifespan through the analysis of particular linguistics variables of an English speaker.


Introduction
There are a number of theories as to how language change occurs as well as how linguistic variation is related to speaker age.Sankoff (2004) discusses four relevant models: generational change (or "apparent time"), agegrading, communal change, and lifespan change.The apparent-time hypothesis was pioneered by Labov (1978:277) to "use the present to explain the past".Labov's theory relies on the idea of the critical period, according to which there is a limited amount of time after language acquisition in which a person's vernacular can change.If this theory is correct, a person's vernacular in adulthood would always reflect what language was like in their community when they were younger-you could gather together people of different ages, and differences in their speech patterns would represent how the language in the community has changed over time.Labov (2006) tested out his theory in New York City, looking at the use of post-vocalic /r/.The younger participants made more use of this /r/ than the older participants, which Labov theorised to show that the community was increasing their use of /r/ over time.
Age-grading is in direct contrast to the apparent-time hypothesis.Wagner (2012:373) defines age-grading as "the instability of an individual's use of a feature over the lifespan against a backdrop of community stability for the same feature".In other words, in age-graded patterns, the rate of use of a stable variant is fixed within a certain age group.Wagner (2012) discusses how non-standard variants are typically associated with younger and older speakers, but not middle-aged speakers.This yields an age-graded curve, where the use of standard forms peaks at middle age.From this perspective, a gathering of people of different ages and speech patterns would be interpreted differently than in the apparent-time hypothesis: the patterns would indicate changes for individual speakers, but not the community as a whole.The apparent-time hypothesis relies on speakers being unable to change their language after the critical period.
Communal change refers to all the individuals in a community changing their use of a linguistic feature simultaneously.Meyerhoff (2011, cited in Wagner 2012:374) illustrates such change with the example of a community adopting a taboo on a deceased person's name.In contrast, a study by Arnaud (1998) on the use of the possessive (ING) suffix found that while the community changed, not all speakers adopted the new variant at the same time.Since it was not simultaneous enough to be communal change, these results were treated as a different type of age-grading.However, the definition of age-grading generally concerns a stable variable, not an unstable change in the community.Therefore, Sankoff (2005Sankoff ( :1011) ) coined the new term "lifespan change" to indicate a "change over [an individual's] lifespan in the direction of a change in progress in the rest of the community".
Using video interviews, we attempt to see if David Beckham has changed his language over the past 20 years, and if so what that says about the theories surrounding language change across the lifespan.We will look specifically at the consonantal variables of h-dropping and t-glottalisation in David Beckham's speech.

Speaker Background
We have chosen to investigate the speech patterns of David Beckham, a former professional footballer, who is most commonly known for his 11-year stint at Manchester United Football Club and for captaining the national squad from 1998-2009.After his Manchester United days, Beckham played for top clubs around the world, including Real Madrid, LA Galaxy Milan, and Paris Saint-Germain, before retiring in May 2013.In addition to his sporting success, Beckham has extended his fame through many different channels.According to Forbes Rich List, Beckham was ranked number 8 in the world's highest paid athlete list in 2013 and has a current net worth of over $45 million (with $5.2 million stemming from his salary and over $42 million from endorsements, including multinational companies such as Adidas, Pepsi, Armani, and H&M).He has also launched his own fragrance line and video games, and has made film appearances in movies such as Bend it like Beckham and The Goal! Trilogy (Bio 2015).
David Beckham was born on 2 May 1975 and grew up in Leytonstone, East London, a suburban area in the borough of Waltham Forest.He came from a working-class family: his mother was a hairdresser and his father a kitchen fitter (IMDb n.d.).The East London dialect is commonly referred to as Cockney, and is said to be traditionally a variety of the working class (Hughes et al. 2013:75).For the purposes of this paper we assume that Beckham's childhood variety was Cockney.Common marked features associated with Cockney include glottal stopping, l-vocalisation, the dropping of /h/, and th-fronting.
During the period we examine in our paper, Beckham's social circumstances and socioeconomic status changed dramatically.The first interview, when Beckham was an 18-year-old, saw him as a working-class teen; in the later interview, Beckham was a 39-year-old multi-millionaire.In the course of his professional career, he has also been highly mobile geographically.He played for a range of teams across the world, depending on which football club he was employed at.Although he and his family have always had a home in London, they have also resided in Madrid, Milan, and Paris.They made a more permanent home in Los Angeles, California, from 2007-2012, but during this time, Beckham had short spells on loan in Milan.His two main places of residence have been Los Angeles, USA, and Kensington and Chelsea, London.
The recordings we analyse in this paper were retrieved online from interviews made available on YouTube.Although there were very few interviews during his teen years to sample from, we collected data from a large selection of material of Beckham in his older years, resulting from his success and sharp rise to fame.

T-glottaling
T-glottaling is the realisation of word-final and word-medial /t/ as a glottal stop.Fabricius (2000) suggests that it originated in London and has rapidly spread outwards, although Schleef (2013) found that there is a longer history of t-glottaling in Scotland than in London; in fact, teenagers in Edinburgh use the variant more regularly than teenagers in London.T-glottaling may originally have been described as a change from below, associated with the vernacular of lower-class adolescents (Milroy et al. 1994), although there is evidence to show its acceptance in Received Pronunciation (RP) in recent years (Fabricius 2000), and it appears to have risen above the level of awareness today.Smith (2005:50) describes RP as "a prestigious accent of English associated, but not restricted to, the South-East of England.It is a standardised norm towards which speakers tend".Smith goes on to state that individuals tend to conform to RP to different degrees, and we expect this to be particularly true for speakers in the South-East of England and/or East London.There are a number of social and linguistic constraints on the use of t-glottaling in English.The most studied environment of t-glottaling is post /t/, where Fabricius (2000) found that pre-consonantal t-glottaling is already accepted in RP, and pre-pausal and prevocalic /t/ are not far behind.The use of t-glottalisation is also subject to style-shifting.Fabricius (2000) found that there was a very significant increase in the use of t-glottaling when participants took part in an informal interview, compared to the more formal reading of a passage.
T-glottaling is associated with "urban" areas and Williams and Kerswill (1999) found it to be most common with young working-class speakers.Due to the association with this particular group of speakers, there have been many questions relating to whether this is a feature of age-grading or a sound change.Pronunciation is commonly thought to be fixed after adolescence; however, factors such as geographical location and social mobility can directly influence our speech.Williams and Kerswill (1999) also found that region affected the rate of t-glottaling, with London and the Home Counties using the variant significantly more than the rest of the UK.Wells (1994) also observes that intervocalic t-glottaling, which is rarely seen across the UK, is a very common occurrence in Cockney English.

H-dropping
H-dropping is the deletion of word-initial /h/ in English.Ramisch (2010) theorises that the establishment of hdropping has been relatively recent in history.Although widely spread in the UK, h-dropping is rare in the US.It must therefore have developed sometime after the English settled in America in the 17 th and 18 th centuries.There are a number of common words in which h-dropping is incredibly prevalent.This group includes the majority of auxiliary verbs (e.g., had, have) and pronouns (e.g., he, her).Drummond (2010) notes that often researchers will discount these tokens altogether, as the likelihood of h-dropping is so high.
There are a number of social constraints on the use of h-dropping.Trudgill (1974) found that h-dropping was more common amongst the working class than higher socioeconomic classes, while Williams and Kerswill (1999) observed that h-dropping was the norm for older speakers in their study of Milton Keynes, Reading, and Hull.Upton and Widdowson (2006) note that most regions of England show some dropping of word-initial /h/, with the exception of large areas around Northumberland and East Anglia, which retain the /h/.

Methodology
David Beckham was our chosen candidate for this paper as he was renowned for his Cockney accent.We decided that it would be interesting to analyse the language choices of an individual who grew up with such a strong Cockney accent, and then found fame, to find out whether that socioeconomic climb influenced his speech.As interviews with Beckham during his teenage years were very difficult to find, we were considerably limited in our choice.Hence, we selected two particular interviews which contained the longest clips of Beckham's speech, rather than that of the interviewer or host.Beckham's first interview (spkmnerth999 2013), which was taken in 1994 at the age of 18, is a media training interview, in which the topic of conversation is a recent match and Beckham's role in the game; unfortunately, that is all that we know about the context of this interview.Given that the interview was a training exercise for the aspiring star, we would assume that Beckham felt relatively at ease as he was not under any obvious pressure.We believe that due to the low-pressure situation, the interview is likely to consistently present Beckham's vernacular.The second interview, conducted by Gary Neville, is with a 39-year-old Beckham on the topic of his decision to officially retire.It was published on 16 May 2013 (Sky News 2013).Once we had selected our interview clips, we converted them using Audacity (n.d.) to a file type compatible with ELAN (n.d.) and Praat (Boersma and Weenink 2016).We transcribed David Beckham's speech in the interviews in ELAN.
Once we had transcribed the interview, we saved it as a tab-delimited text and input the file in FAVE-align (Rosenfelder et al. 2011), along with the sound file from the interview.We modified the pronunciation dictionary to include any unidentified words, resubmitted the transcription, and modified the dictionary.FAVEalign then produced a text grid file for use in Praat.In order to identify our tokens, we used a Praat hand coder script (Fruehwald 2011).This was a much more efficient way of collecting our tokens rather than doing so manually.The hand coder allowed us to choose variables and the environments we wanted to look at, and then to select each token from the text grid.
For t-glottaling, we excluded tokens followed by /t/, which would have made it difficult to accurately code whether the /t/ was present or glottalised due to the influence of the following segment.We also decided to exclude high frequency tokens such as it which we thought could influence our results.Tokens which were blocked from t-glottalisation due to stress patterns were also excluded, such as the words into, country, and fifteen.
While coding for h-dropping, we decided to only code word-initial /h/ and exclude the pronouns him, her, and his.After we had deleted these, we found that the 1994 interview only yielded 8 tokens of word-initial /h/.We decided that this would not offer us enough data to draw solid conclusions from our results, so we found a second interview from 1994 to transcribe and code.
Before analysis, we examined each token in Excel to check for errors produced in the initial output file.We ensured that we had excluded words which were outside of the envelope of variation, for example, deleting tokens such as country, whose /t/ cannot be glottalised in the stressed position.This was an important step as the hand coder showed all tokens of /t/ followed by a vowel or a nasal in intervocalic position, but we had to make the judgement as to whether /t/ could be glottalised in that particular token.

Hypothesis
Our hypothesis for the investigation was that David Beckham's rates of variable use would change over the 20 years between the interviews analysed.Specifically, we were expecting to see a significant decrease in both tglottaling and h-dropping.T-glottaling and h-dropping are stigmatised forms commonly associated with a working-class background.Due to Beckham's dramatic upward social climb, we would expect his language choices to reflect his current socioeconomic status.This means that Beckham is likely to use the prestigious standard forms (/t/ and /h/) more frequently in his later interview due to the effects of interview training and the change in his social circumstances.For /t/ production, in particular, we expected to see especially high rates intervocalically, due to this environment being the most salient and stigmatised when a glottal stop is present.

H-dropping Results
Beckham has massively reduced his use of h-dropping from 100% of the time in 1994 to only 33% in 2014 (see Figure 1).The 100% use at age 18 is as expected.Beckham's social and regional standing conformed with what we already knew about h-dropping.Although present across the UK, h-dropping is often associated with Cockney English, used in the East of London where Beckham grew up.With his parents' socioeconomic situation, Beckham was born into a working-class family.H-dropping is traditionally seen as a working-class feature and is stigmatised as being "uneducated" (Ramisch 2010).In fact, a London dialect labelled "Estuary English" has developed, which Ramisch (2010:175) defines as "an intermediate variety between the most localised form of London speech (Cockney) and a standard form of pronunciation".In this dialect, h-dropping is specifically avoided because of the stigma it holds and its association with lower-class speakers.This, along with evidence from Bell and Holmes (1992) that men h-drop more regularly than women, provides a strong case as to why Beckham h-drops so regularly in his early interviews.The rate of 39-year-old Beckham's use of h-dropping also supports our hypothesis.As discussed, h-dropping is closely related to social standing, and Beckham has undergone a huge social climb in the 20 years between interviews.With an estimated net worth of $350 million and an OBE awarded by the Queen, Beckham's family now enjoy high socioeconomic status and may therefore be less likely to exhibit h-dropping.Between the first and second interviews, Beckham also relocated to Los Angeles.H-dropping is not a linguistic variable often found in America (Ramisch 2010), and certainly not in Los Angeles.Therefore it is possible that Beckham's reduction in h-dropping is also the result of his having undergone dialect convergence, exhibiting some features of American English.
Beckham's age in his 2014 interview might also be a big factor in the results we see.As previously discussed, language change across the lifespan often follows a curvilinear pattern, with non-standard forms being used most frequently in adolescence and older age.Meanwhile, middle-aged speakers, as Beckham is in his 2014 interview, are typically more likely to use standard forms, which in this case would be the retention of /h/.Our results are in line with the model of age-grading, as Beckham exhibits instability with a variable that is stable within the community.
We also found an interesting correlation between the preceding segment and rate of h-dropping.Although the number of post-pausal tokens were to too few to test statistically, we found a significant difference between post-vocalic and post-consonantal /h/ (p < 0.05).There seem to be few findings from previous studies on the linguistic restraints conditioning h-dropping, with the majority of research focusing on social context.Although we looked at various environments, such as word class and the position of the following vowel, neither rendered significant results.We suspect that there are no other linguistic restraints on h-dropping aside from the preceding segment type, which would support the dearth of previous studies.It may also be that our results in favour of post-vocalic /h/ are unusual, and we would have to code more /h/ tokens at a later date to ensure our results are correct, as there were so few of them.
Overall, the results we obtained from our analysis of David Beckham's use of h-dropping confirm our hypothesis that his use of the stigmatised form would have decreased over time.Our study also supports the age- grading theory of language change, providing evidence that it is very possible for language to change across the lifespan, even after the critical period has passed.

T-glottaling Results
Our findings for t-glottaling are again in line with our hypothesis that Beckham would decrease his use of the glottal variant in all word positions and phonotactic environments.In the first interview, as with h-dropping, he glottalised /t/ 100% of the time.In his later interview, he only uses the glottal variant 82% of the time.Again, a drop is as we expected, given the stigmatisation of the glottal variant: Beckham has experienced a dramatic social climb and his language choice reflects his higher socioeconomic status.Moreover, age could also be a factor.T-glottaling is more frequent in young speakers, and Williams and Kerswill (1999) state that it is one of the features of a set of youth norms, with older speakers using glottalised /t/ to a lesser extent.Since t-glottaling is also a Cockney dialect feature, it is not surprising that Beckham exhibits a 100% glottalisation rate in his first interview.
The 18% decrease in the glottal variant by Beckham in his 2014 interview is less of a drop than his change in h-dropping, and yet we maintain that it constitutes a major reduction.As seen in Figure 3, the change is actually much larger intervocalically than word finally.Beckham went from using the glottal variant 100% of the time in the first interview to just 57% of the time by the second interview, which is a 43% decrease.After testing the significance of the decrease via a chi-square test, we got a p-value of 0.03, which allows us to confidently state that this is a significant difference over the two interviews.Beckham demonstrates a decline in both word-final and intervocalic environments; however, the largest difference comes from the intervocalic environment, where he drops from glottalising 100% of his tokens to just 57%, which is a 43% reduction.In word-final position, Beckham glottalised 83% of his tokens at age 39, which is a 17% reduction over his interview at age 18.
Although both environments show a decline in the use of the non-standard variant, the intervocalic position has undergone the largest change.Intervocalic /t/ is extremely salient and carries much more stigmatisation than word-final /t/.Wells (1997, cited in Fabricius 2000:119) states that t-glottaling can be used in RP in some environments; however, t-glottaling intervocalically and before syllabic /l/ has not been cited as a feature of RP.This suggests that intervocalic t-glottalisation is still viewed as a less-favourable variant, which lacks the prestige of the standard /t/ variant.While it is likely that Beckham has become more conscious of his language choices and needs to adhere to the standard more often, our results may also suggest that he feels it is acceptable to glottalise in the word-final position, which is why his rates of the glottal variant are still considerably high.In contrast, he may be avoiding using glottal stops in the intervocalic position given his age and higher socioeconomic status in the 2014 recording.

Effect of Following Segment on Glottalisation
The segment type following /t/ also influences the variant Beckham uses.The results show that he, again, glottalised 100% of the time in both environments in 1994.Twenty years later, however, we see a large reduction in glottals in the position of word-final /t/ preceding a vowel.This has dropped from 100% to 14%, a decrease of 86% (p < 0.001).The tokens in a medial position, when followed by consonants, have remained at 100%.This intermediary glottalisation is a salient variant which is classed as a working-class feature.Our data 0% 20% 40% 60% 80% 100% 18 (1994) 38 ( 2014)

Intervocalic
Word Final therefore support the description by Schleef (2013:204) that the pre-consonant environment is where the glottal variant most often occurs, with pre-vowel environments disfavouring the glottal.Fabricius (2000:141) found that in the word-final position, glottaling before a consonant was already accepted in "modern RP".She also predicted that glottaling before a pause would become an accepted feature of language in the next generation.Given that the second interview recording took place 14 years after her work, we could argue that Beckham's language change is due to the glottal variant being viewed as more acceptable in the most modern forms of prestigious dialects.We see a large reduction in glottals when a vowel follows, in the position of word-final /t/ preceding a vowel %; on the other hand, the tokens in a medial position, when followed by consonants, have remained at 100%.(Please compare Figures 4 and 5.)

Effect of Preceding Segment on Glottalisation
The highest rates of glottalisation occur when the /t/ is preceded by a vowel.Aged 18, Beckham glottalised 100%; aged 39, he glottalised 73% in words where /t/ was preceded by a vowel and 29% when they were preceded by a consonant (see Table 1).His glottaling rates are lower for intervocalic tokens than word-final tokens, which again reinforces our hypothesis that the rate of t-glottaling has generally decreased, particularly in the salient intervocalic environment.Overall, the results clearly show that there has been a statistically significant change in David Beckham's production patterns of /t/ versus the glottal stop over the last 20 years.As discussed, we think this is due to the change in his social circumstances, both in terms of social and geographical mobility.

Conclusion
Our investigation has supported our hypothesis that David Beckham's language has changed between the two points in the 20-year period which we examined.The most surprising result was the change in h-dropping, with Beckham opting for the standard form 66% more frequently.Though Beckham's rate of t-glottaling is still relatively high, he has decreased its general use by 18% overall and 43% intervocalically.While we have outlined that the change in Beckham's choice of variants is in line with what we would expect of someone of his socioeconomic status, we do not suggest that this is the only factor at play.It is important to consider a range of social factors, including age and geographic mobility, such as when Beckham moved to California to play with the LA Galaxy Football Club.The 20-year period between the interviews sees Beckham at two very different life stages: as a teenager and as a middle-aged, successful, and well-established man.It is also important to outline that by the second interview, Beckham has already had a long career in the public eye and gained extensive experience with interviews, ranging from talk shows to post-match debates.That is experience that the 18-year-old Beckham did not have, whose speech is thus more natural than that amended for interview purposes.
If this study were to be repeated, it would be beneficial to find sources containing more speech to increase the number of available tokens.Of the two interviews combined, there were fewer than 200 /t/ tokens, which we have treated as an acceptable data sample; however, working with double the amount would have given us more reliable insights into Beckham's change in glottalisation 20 years on.There were very few /h/ tokens, making it difficult to analyse the linguistic factors conditioning h-dropping.Had we had more tokens, a more in-depth analysis could have been possible.In addition, when coding /t/, it might have been useful to exclude words in consonant clusters when preceded by plosives or fricatives, such as "first" (Schleef 2013), which we did not do here.
Our research into David Beckham's consonantal variation across his lifetime has confirmed that an individual's language does change across their lifespan, and this can be the result of a range of social factors, including age, socioeconomic status, and geographic location.Analysing the language of a high-profile individual adds the further dimension of being regularly scrutinised in every aspect of their life, including the very way that they speak.

Figure 2 :
Figure 2: Beckham's use of h-dropping at age 39, by preceding segment type.

Figure 3 :
Figure 3: Beckham's rates of t-glottaling at two different ages with respect to position in the word.

Figure 4 :
Figure 4: Beckham's rate of t-glottaling aged 18, by word position and following segment.

Figure 5 :
Figure 5: Beckham's rate of t-glottaling aged 39, by word position and following segment.

Table 1 :
Beckham's rate of T-glottaling at age 39, by word environment and preceding segment