Forensic Linguistics Works

Index of articles

Return to main index of the Forensic Linguistics Works

Short Analysis of a Saudi confession

This is the 'confession' spoken by Alexander 'Sandy' Mitchell on Saudi Arabian television in 2001:

Text 1: Mitchell 'confession'

"My name is Alexander Hutton Johnston Mitchell. I am also known as Sandy. I am of British nationality and I work at the Security Forces Hospital here in Riyadh.

I am the chief anaesthetic technician. I confirm and confess that I was ordered to carry out an explosion here in Riyadh, which took place on Friday, 17 November 2000. The explosion was directed against Mr Christopher Rodway, who is of British nationality.

During this explosion I was assisted by Dr William Sampson, of Canadian nationality. I placed the explosive device under the driver's seat of Christopher's car. That afternoon, Christopher came out with his wife, got into the car and drove off. William detonated the remote control, which caused the explosion in Christopher's car.

We then turned south away from the scene of the incident. We found out later that Christopher had been killed in the explosion and his wife had been injured.

A second explosion was authorised and we carried out the order in a car which was similar to that of Christopher Rodway.

The second explosion was planned for 22 November. As Bill and I were discussing the results of the first explosion, a friend of ours, Mr Ralph Carl, a Belgian citizen, overheard our conversation. It became necessary for us to involve Ralph in carrying out the second explosion to ensure his silence."

In this case I found it interesting to look at some of the non-native features of the text. At the same time I decided to combine these with sentence length. I was interested to see whether any sentence lengths in the text differ significantly from the average. In other words how great is the standard deviation.

Graph 1: Mitchell text - sentence length

From the graph we see that sentence numbers 5, 13 and 15 appear much longer than the rest of the text. As noted above the sentence length mean for the text is 13.1 words. The amount by which sentences deviate from this, on the whole, is 5.88. This (informally) is the standard deviation for the text. As a very general rule of thumb when any individual sentence differs by more than twice the standard deviation from the mean, then this sentence is said to be 'statistically significantly different' from the rest of the sentences in the text. In fact sentences 5, 13 and 15 do deviate by approximately 2 standard deviations from the rest of the text. Below the text is presented again, this time with the very long sentences in italic, bold type:

Text 2: Mitchell confession presented as numbered sentences. Sentences approximately (or more than) two standard deviations from the mean are given in bold, italic type.

1. My name is Alexander Hutton Johnston Mitchell.
2. I am also known as Sandy.
3. I am of British nationality and I work at the Security Forces Hospital here in Riyadh.
4. I am the chief anaesthetic technician.
5. I confirm and confess that I was ordered to carry out an explosion here in Riyadh, which took place on Friday, 17 November 2000.
6. The explosion was directed against Mr Christopher Rodway, who is of British nationality.
7. During this explosion I was assisted by Dr William Sampson, of Canadian nationality.
8. I placed the explosive device under the driver's seat of Christopher's car.
9. That afternoon, Christopher came out with his wife, got into the car and drove off.
10. William detonated the remote control, which caused the explosion in Christopher's car.
11. We then turned south away from the scene of the incident.
12. We found out later that Christopher had been killed in the explosion and his wife had been injured.
13. A second explosion was authorised and we carried out the order in a car which was similar to that of Christopher Rodway.
14. The second explosion was planned for 22 November.
15. As Bill and I were discussing the results of the first explosion, a friend of ours, Mr Ralph Carl, a Belgian citizen, overheard our conversation.
16. It became necessary for us to involve Ralph in carrying out the second explosion to ensure his silence.

Of these sentences, sentence 5 is the longest. What is most striking about this sentence - apart from its length - is the alliteration: 'confirm and confess'. This phrase is very curious, as such blatant examples of alliteration are generally avoided by native speakers. It appears possible that the phrase could be by a non-native speaker. Sentence 13 appears to contain a possible error, namely 'we carried out the order in a car which was similar to that of...' If the order was to carry out an explosion, then it would seem that the order would be carried out 'on a car', rather than 'in a car'. Indeed if the order had been carried out in a car, the narrator would probably not be alive to describe what had happened. This kind of distinction ('in a car' vs 'on a car') is absolutely fundamental to what native speakers know about their language.

Also, Christopher Rodway should be apostrophised: '....a car similar to that of Christopher Rodway's', though the phrase is clumsy and would have been better as '....a car like Christopher Rodway's...' (Incidentally, it is worth noting that the victim, Christopher Rodway is referred to in no less than three different ways in this short text of 226 words: 'Mr Christopher Rodway', 'Christopher', and 'Christopher Rodway'). There is another particularly interesting non-native speaker like feature of the text, and that is the phrase 'I am of British nationality'. This is a very official, formal phrase, and not the kind of phrase someone would use except perhaps when referring to someone else in an official capacity. We would probably say: 'I am British' rather than 'I am of British nationality'. It is a very official phrase, and I believe a native speaker would not make the mistake of mixing registers like this. The grammar is not at fault, but the idiom is certainly 'heavy'.

However, despite these non-native speaker like features, there are several pointers towards a native speaker having composed the text. Some years ago I conducted some authorship tests on a set of texts about discourse analysis from a university text book. As it happened, some of the texts were by non-native speakers, though it would not have been possible to guess this from just reading the texts. However, I noticed that non-native speakers tended to exhibit lower lexical density values than the native speakers. Laviosa (1998) has noticed something similar, and reports, in connection with translated articles that regardless of the source language these texts exhibited lower lexical density, lower word length average and lower sentence length average than non-translated texts. I am not sure whether this is relevant to the discussion of native vs non-native speakers or not, but in any case the Mitchell text exhibits high values: high lexical density (50%), and a high word length average (4.87 letters per word). These values would not suggest, at first sight, a non-native speaker. Nevertheless, we cannot ignore such strange phenomena as 'confirm and confess' and 'in a car' rather than 'on a car'.

These observations led me, after careful consideration, to consider the possibility that the Mitchell text had been authored by a native speaker and a non-native speaker working together. It would seem that the non-native speaker was the more powerful influence of the two, otherwise presumably the lacunae referred to above would not have passed the scrutiny of a native speaker. Short as the text is, it does appear to contain two styles.

There are many ways of two people authoring a text together, and the combination of an interrogator and a native speaker having to co-operate to produce a text can throw up some interesting lexical combinations and idiomatic paradigms. The phrase 'confirm and confess' not only exhibits an unusual example of alliteration, it also re-inforces an institutional perspective, which is that of emphasising culpability. It therefore has echoes of 'police register'.

References

Laviosa, Sara. 1998. 'Core Patterns of Lexical Use in a Comparable Corpus of English Narrative Prose', in Sara Laviosa (ed.), 557-570.
Laviosa, Sara (Ed) Translation studies abstracts, Vol 1, number 2, 1998 Manchester, St Jerome, 1998.