Profiling for the media

I was recently called upon by Associated Press, the BBC News Channel and Reading's Jack FM to comment on the spoken features of the jihadist from the video footage of the beheading of US journalist James Foley, as it has been suggested that he is British.  You can see me in the Associated Press clip on French television below.

I'm always ready to be called upon to comment on linguistic issues for the media, but this situation was particularly difficult owing to the content of the video.  I hope that the phonetics community is able to assist the authorities. I wish to make it clear that am not carrying out the analysis myself.

The speaker displays many of the features of a British accent known widely as Multicultural London English (MLE), such as producing vowels in e.g. FACE and PRICE as monophthongs, not dropping /h/ sounds (/h/-dropping is common in London accent Cockney) and pronouncing voiced dental fricatives in e.g. the as a [d].  There are glottal stops, which are less common in Afro-Caribbean or African English accents, and /l/-vocalisation. The speaker also has a more syllable-timed speech rhythm; instead of pronouncing the phrase from all walks of life as /frəm ɔːl wɔːks əv laɪf/ it sounds more like [frɒm ɔː wɔːks ɒv lɐːf], with a full vowel in each syllable.

See HERE for information on the features of MLE (yes, it's Wikipedia, but a good summary).

The accent was identified chiefly by Professor Paul Kerswill and colleagues; Paul was at Reading, but is now at York via Lancaster.

It's impossible to say exactly how many speakers there are of this accent, but it is common among younger working-class speakers in the London area, and features of the accent have also been observed in other urban areas of the UK.  It is not at all exclusive to speakers from an Afro-Caribbean background but is also spoken extensively but e.g. white and Asian speakers wishing to identify with a certain demographic / social group.

British impressionist and actor Alistair McGowan did a nice piece on MLE for the BBC's One Show, which you can view below.

I would say that the speaker in the clip is probably a UK or fully bilingual speaker of English rather than a second language learner or someone with an indigenised variety of English (e.g., Nigerian English). The speaker probably grew up in or near inner London and has probably been educated in the UK system.  I would be surprised if he was from outside the greater London area, but this is an accent which is socioculturally attractive and so he may be from further afield. I would also suggest that he is lower middle-class rather than working-class as he sounds educated.

It should be noted, however, that we cannot actually see him speak in the film. Most of his face including his mouth is covered.  It could, therefore, be a voice-over.

When I appeared on the BBC News Channel (I'm so sorry I don't have a clip of this to share) I was asked about forensic phonetic analysis of this speaker's voice. What we would need to be able to do this is a reference sample of a known speaker in order to make comparisons between that and the Foley video.  As one of my colleagues, Martin Barry, points out, unless this speaker has spoken into a police microphone it will be almost impossible to carry out forensic speaker comparison successfully.

My picture from the AP session also appeared in the Los Angeles Times.  You can view the online article HERE, which has comment from Martin Barry.

Phonetic vs phonemic inventories

In my first year "Sounds of Language" class, one of the things we do is look at phonetic vs phonemic inventories. I've just had a question about this on the discussion board for the module so I thought I may as well post my response, in case anyone is interested.

Sounds pattern differently in different languages. Speakers of two languages may produce exactly the same set of speech sounds - or phones (phonetics) - when talking, but the languages may use those sounds differently to create meaning. Once we're talking about meaning, we are considering phonemes.

Say, for example, there are two languages whose speakers produce the consonant sounds [p] and [b] and have one vowel, [a]. In both cases, the phonetic inventory contains [p], [b] and [a]. We put the sounds in [] brackets to indicate we are just talking about how the sound is produced at the moment. 

The only thing which is different between [p] and [b] is voicing; [p] is voiceless and [b] is voiced. Otherwise, they are both bilabial plosives.  

In language A, [bapa] and [baba] mean different things - [bapa] means "red" and [baba] means "yellow". [bapa] and [baba] constitute a MINIMAL PAIR, as only one sound differs between the two and it changes the meaning of the word. We can therefore say that /p/ and /b/ are phonemes - meaning units - because of this change in meaning, and we now put them in // brackets.  There are TWO consonant phonemes.  Thus, the phonemic inventory is /p/, /b/ and /a/.

In language B, however, [baba] and [bapa] both mean the same thing - they both mean "car". This means that it doesn't matter whether the consonant is voiced in language B. As there is no change in meaning when one substitutes [p] for [b], they are NOT different phonemes but belong to the same single phoneme. 

What we have to do for language B is decide which sound represents the phoneme, and we often choose the one which occurs in most environments. As we don't have a lot of data here, let's go with the phoneme being /b/ (as there are more of them). That phoneme contains the two sounds [p] and [b], which are ALLOPHONES (phonetic variants) of /b/. Thus, the phonemic inventory is /b/ and /a/. 

This is a very limited set of data, however!

The Guess List: a study in /t/ elision

The BBC is airing a new game show on Saturday nights hosted by the wonderful Rob Brydon and amusingly entitled The Guess List. You can read the Independent newspaper's less than glowing review of it here.

Why amusing?

This plays on a phrase, the guest list, which is a list of people invited to an event, i.e., a list of guests. No surprise there.

What amuses me is that it is an example of how the process of alveolar plosive elision can result in homophones in English - in this instance, a homophonic phrase.

When one produces the phrase the guest list in rapid speech, it is normal to leave out the /t/ sound at the end of /ɡest/.  This process is called /t/ (or /d/) elision.  The rules for when this can take place are as follows:

1. The /t/ or /d/ must be in the syllable coda;
2. It must be surrounded by other consonants;
3. The consonant preceding the alveolar plosive must agree in voicing with it - so if the plosive is a /t/ it must be preceded by a voiceless consonant and, if it's /d/, it must be preceded by a voiced one.
4. The consonant following cannot be /h/.

So, in guest list, which can be transcribed phonemically as /ɡest lɪst/, we can elide the /t/ at the end of /ɡest/ because it meets the requirements listed above.  This results in /ɡes lɪst/, which means guest list and guess list are homophonous.

There is, as far as I know, no such thing as guess list as a phrase in English. If one types it into Google, for example, it redirects you to guest list.

Other notable examples of homophones resulting from connected speech processes include handbag /hændbæɡ/ becoming homophonous with ham bag /hæmbæɡ/. There are two processes going on here: /d/ elision and assimilation.

Assimilation is a process by which sounds at word boundaries - often alveolar consonants - become more similar to each other in rapid speech. Here's a diagram showing consonants at word boundaries:

_ _ Cf | Ci _ _

Cf = final consonant; Ci = initial consonant

In English, we tend to get regressive assimilation, which means the initial consonant (Ci) at the beginning of the next word has a backwards effect on the final consonant (Cf) of the preceding word. As I mentioned above, this tends to affect alveolar consonants, and more often than not it will affect the place of articulation of Cf, i.e., it will not be produced as an alveolar consonant but will have the same place of articulation as the Ci of the next word.

In handbag /hændbaɡ/, the alveolar plosive /d/ is elided and the alveolar nasal /n/ is produced as a bilabial consonant because the following word - bag - begins with a bilabial consonant, /b/.  This results in the production /hæmbæɡ/, which is homophonous with ham bag. But of course, a lady wouldn't normally take a bag of ham out with her when she went shopping, and we can usually retrieve the real meaning from the context.

I should add a caveat in that this is a very brief overview of the theory of these two processes. In very rapid speech, all sorts of sounds get elided and / or assimilated, so analysing spontaneous speech can be a real challenge.

Another issue which arises from connected speech processes such as elision and assimilation is that they can make speech less intelligible or the message more difficult to understand.  Emilio's comment below led me to this example, spoken by Elizabeth Taylor and Richard Burton in Who's Afraid of Virginia Woolf? In it, there is /t/ elision in the word guests, which is perfectly legal. You can see how Burton's character doesn't understand Taylor's character until she repeats the word guests with the /t/ in it - although what "We've got guess" (as opposed to "We've got guests") might mean is difficult to ascertain.  There are obvious issues for speech intelligibility here in English as an international lingua franca.

Watch from 04:04 right near the end.  And thank you, Emilio!

I'd recommend the following books by way of introduction if you are interested in connected speech processes in English:

Lecumberri, M L G & Maidment, J. 2000. English Transcription Course. London: Arnold.

Roach, P. 2009. English Phonetics and Phonology: A practical course. Cambridge: Cambridge University Press.