Written and Spoken Utterances as Special Evidence
Spoken or written threats, suicide notes, confessions, declarations in wills, and a range of other utterances may become part of civil or criminal proceedings or the associated investigations. These reports of what people have said or copies of what people have written may be examined in order to answer a number of different questions: for example, exactly what the utterances mean; what is likely to happen as a consequence of them; what the characteristics of their author are likely to be; who the author is likely to be; or whether they are genuinely the product of the person who it is claimed produced them.
The value of studying words that are written as part of a crime is that text does appear to offer special potential as evidence. A threatening extortion letter is a record of exactly what the offender did to commit the crime. The letter is the crime scene. There is no need to make sure the witness saw what was happening or to puzzle over the pathologist’s report. On the page is a complete record of the relevant criminal actions.
Similarly, a note left declaring the reasons for a suicide is a valuable piece of evidence, if found with a dead body. It may help to show what the person was thinking and feeling and thus lend support to any belief that there were no other culprits. It is a very special kind of record of inner despair that has no equal in observations of what the person did before death, or even what others report he or she had said.
The written confession is also held in high regard by the courts and public alike. It is the fact that culprits have described in their own words the actions that incriminate themselves that is so significant. Their words are seen as providing direct contact with their guilt.
There are many other aspects and types of crime in which some form of written record of the words of the offender exist. Stalking may be one such, in which offensive letters may play an important part. Various types of business fraud may also leave traces of the writing of the offender. All these written accounts offer a direct glimpse into the mind of key actors, if we can find a way to interpret those glimpses.
In some cases the questions are about the usual or particular meaning and usage of the words. An example would be whether the utterance was likely to be interpreted as a genuine threat by the person who received it, or whether the text implies connection with a brand in a trademark dispute. In these cases the expertise required to answer the questions draws upon knowledge of the language in question and would be provided by a linguist, typically one who has taken a particular interest in the forensic application of linguistics.
Psycholinguistics Distinguished
When the questions about written or spoken material deal with aspects of the characteristics, behavior, intentions and mental state of the author, the expertise involved is more likely to draw upon that branch of psychology that deals with linguistic issues, often known as psycholinguistics. Most notably, for instance, is the consideration of whether a suicide note is genuine, as might be part of a psychological autopsy. However, the overlap between the different disciplines and pseudodisciplines that have an interest in this area is quite considerable. Often differences in expert opinion on the same verbal material will be a consequence of differences in the discipline that is brought to bear, rather than some substantive difference in actual measures or findings. Because of these potential confusions some more clarifications are of value before we turn to consider the modest contribution that psycholinguistics can currently make to forensic matters.
The first clarification to make is that the pseudo-discipline of ‘graphology’ has no place in the forensic process. Graphologists claim to be able to interpret the character and mental state of an author from the way he or she shapes letters on paper when writing. Graphology therefore focuses on the actual physical way in which the author writes and how they arrange the words on the page, in order to make psychological inferences about the writer. Although graphologists are widely used in industry, especially in France, for selection of personnel, and there are many anecdotes about their contributions to police investigations, every thorough systematic, scientific study has failed to produce evidence to support the claims of graphology. Furthermore, graphologists claim they must know the writer’s age, sex and nationality before they can comment on the personality because none of these facts is revealed by the writing. For many investigations, of course, these crucial facts are what the investigator wants to know.
There are also some basic problems in using the analysis of handwriting: although handwriting does consists of measurable elements, such as slant and size, and descriptive elements, such as letter form and tendencies to the right and left, there has been little progress in establishing a common denominator for any of these elements to permit comparable objective evaluation. So although there is still a scientific interest in the possibilities that graphology may offer, there is no consistent evidence of its validity. Nor is there any psychological reason to expect that personality should be revealed through the ways in which written letters are formed. And, of course, in many situations criminals deliberately avoid writing. The widespread use of word processors is doubtless further reducing the impact of this pseudoscience.
Graphology should not be confused with the work of handwriting analysts. These forensic scientists often have a training in one of the natural sciences, typically physics. Their task is to examine the physical structure of any written material to determine whether it has been modified, or for comparison with other examples, for instance to see if a signature has been forged. They do this in much the same way as fingerprint experts, by determining the salient features of the material in question and establishing what proportion of the features are common to any target and comparison material. It is also important to separate what might be called the physics of writing analysis. Whether it is the shape of a signature or the font of a typewriter, it may be possible to carry out precise analysis to link the physical trace to its originator. But no-one carrying out such analysis would suggest that it could also tell you whether the writer was paranoid or likely to follow through with his or her threats. When the language exists in the form of sound rather than writing, questions about its authenticity or similarity to other material will draw on knowledge of phonetics, requiring the expertise of a forensic phonetician.
Stylistics
Some attempts to determine the authorship of written material have drawn very heavily on numerical procedures. The number of words with more than five letters in the text have been counted, for example, or the frequency with which certain combinations of words occur, such as ‘and then’. There is a tendency to refer to this highly numerical approach in determining whether a piece of text is in the style of one author or the other as ‘stylistics’. The term is sometimes rather loosely applied to any approach to establishing the distinct qualities that determine the particular characteristics of an author.
These numerical methods have become more common with the advent of computers and may draw heavily on the skills of linguists in determining parts of speech and unraveling the structures of language. Historically, however, this area has suffered from being used in contexts in which the actual author of the text is unknown. The procedures are then used to suggest authorship, even though there is no external validation. It is on the basis of such explorations that claims have emerged that it is possible to determine the authorship of a text, or at least whether or not it has multiple authors. However, when carefully controlled experiments have been conducted using texts of known single and multiple authorship, they have been unable to support claims that any stylistic procedure can reliably distinguish single from multiple authors or indicate who an actual author is likely to be.
The most notorious of these ‘stylistic’ procedures is known as ‘Cusum’ or ‘Qsum’, standing for cumulative summation. This procedure, developed by Reverend Quentin Morton, uses arithmetic procedures derived from engineering quality control. In essence it consists of finding the average number of two-and three-letter words per sentence in a piece of text and graphing the cumulative differences from this average across the full text. On to this graph is imposed a further graph, derived from the cumulative differences from the average of the total number of words in each sentence. Morton and his associates claim that any apparent divergence of these two graphs from each other is a prima facie indication that different authors are involved in the creation of the text. Such a curious claim was allowed into a few court cases until systematic research showed that it had no empirical validity.
Contributions of Psycholinguistics
In some cases relatively straightforward psychological considerations of texts can be of great use. For example, it has been found in a number of cases of anonymous, offensive letters that the author reveals so much about his or her own knowledge and intellectual capabilities through characteristics in the way the letters were written that it was clear that the person continually described in the third person could only have been the writer. This type of revelation is more likely when a fluent but poorly educated writer is too arrogant to see how he or she is giving him- or herself away. In most cases the examination of what is written is much more difficult, so much so that a whole new science is struggling to take a first few steps, gleaning whatever can be of investigative value from the examination of written texts.
The questions that forensic psycholinguists are seeking to answer fall into three groups. One set is to do with the identity of the writer. Is this the way in which a known individual expresses him- or herself? Asecond set of questions relates to the character of the writer, really trying to answer the same types of questions as graphologists about the traits and personality of the person writing. The third falls under the heading of prognosis. What is the writer likely to do under various circumstances?
Identifying Authorship
The questions of identity have probably been the most explored and have certainly been the ones that have most exercised police officers. Claims that confessions are not the words of the accused have opened the doors to a series of appeals against conviction and complaints against the police. In these cases the forensic linguist or psycholinguist is claiming that the accused has a unique style of expression and that the confession contains indications that are inconsistent with that style. This draws on the rather attractive notion that we all have a unique style of speaking or writing. The task of the analyst is to specify the details of that style and then demonstrate whether the questioned text accords with those details.
A few studies do show that, under certain, special circumstances, specific indicators may act as ‘signatures’, such as very specialist vocabulary, jargon or slang, unusual spellings or inaccuracies in grammar, or peculiar ways of putting the text on the page, punctuating it or adding emphasis. These may be of use as evidence when two documents are being compared; however, it is difficult to find any examples in the legal record where such linguistic analysis has been accepted without question, although it may still assist investigations.
One example that illustrates this approach is a case in which an incriminating, but anonymous, diary was compared to the known writing of a suspect. A number of misspellings were found in both the questioned and known texts (e.g. ‘breath’ instead of ‘breathe’, ‘its’ instead of ‘it’s'). These also had consistencies in the form of error that could be related to how the words might be pronounced. There were also a number of profanities common to both sets of text, such as ‘ass’, ‘butthole’ and ‘screwed’. Furthermore, there were similarities in the way time was recorded and the forms of expression of emotion. A number of grammatical constructions were also shown to be similar in both samples. It was therefore proposed that the two sets of writing had the same author.
But, even in such an apparently strong example, there are problems in being certain that the indicators really can be treated as ‘signatures’. It is possible, for instance, that most people in the suspect’s circle of associates misspell ‘breathe’ and ‘it’s', that ‘ass’, etc. are common words in their vocabulary and that they all express themselves in a similar way. Without knowing the prevalence, and co-occurrence, of all these constituents in the relevant community, such matches can only be taken as a useful indicator.
Any consistent differences that might be found between the writings of two people may well be a consequence of the language community from which they come. It might be, for example, a product of a particular school, or prison, not unique to an individual.
Even musicologists cannot distinguish all Haydn from all Mozart because both these composers use many of the idioms of their time, as we all do in language.
The problems of knowing the prevalence in the population of any particular aspect of language may make it difficult to use those habits as evidence of identity, but it does not rule out the possibility of narrowing the search for suspects by knowing which language subgroup a writer might come from. Those who study the social differences in the use of language, sociolinguists, have explored the different ways in which people communicate in different subgroups of the population; for example, the larger vocabulary and more complex sentence constructions of educated people and the tendency of professionals to use passive verb forms are widely recognized.
As attractive as the notion is of us each having a unique style of written or spoken expression, or there even being distinct subgroups in the population who have identifiable styles, many processes reduce the possibilities of each person having an identifiable style in all their communications. Verbal communication is a means of contact with others. It is therefore modified in a host of subtle ways to adjust both to the content of what is being expressed, and to whom it is being expressed. Police officers will certainly express themselves in a virtually different language when telling a bawdy joke to friends in a pub compared with when giving evidence in court. We all tend to use much richer and denser vocabulary, and more grammatically correct forms of language, when writing than when we speak. As a consequence, the measurement of a person’s style has somehow to be calibrated to the context in which the communicating is taking place.
Studies have shown, for example, that people use rather different grammar and vocabulary when writing extortion letters compared with when they write personal threatening letters. Some aspects of their mode of writing may carry across the two contexts, but any simple measure of style, such as vocabulary or syntax, turns out to be surprisingly dependent on the nature of the threat being expressed. Much more careful study is therefore needed before we can be confident of clear stylistic differences between people in how they write. This requires the study of the frequencies with which words, groupings of words and grammatical forms occur in different contexts. Developments in computing are now making this possible by the analysis of vast collections of examples of text. But these bodies of material at present tend to be very general, drawn often from newspapers and other published sources. To make a real break- through in understanding texts of interest to the police, we need to work on samples of the relevant types of documents.
At a rather more general level of analysis it may be possible to detect similarities in what a person writes rather than how it is written. This is really an exploration of the person’s ways of thinking rather than the style of expression. Such studies are open to many difficulties in obtaining precise measurement and objective indices, but it has been shown, for instance, that genuine suicide notes have a distinctly different psychological tone to those that are written as simulations for experimental purposes. Genuine notes tend to be longer and more explanatory, indicating more clearly that the author has internalized the decision to take his or her own life.
In a similar form of analysis it has been found possible to determine if two extortion letters are by the same person. They may be typed in very different ways, with intriguing differences in vocabulary, but an examination of the themes in the letters and how they are organized can sometime indicate remarkable similarities. For example, both may start with an instruction to take the threat seriously, followed with a declaration that the threat was feasible, give specific instructions about how the money should be bundled, repeated warnings about not involving the police, and so on. These similarities of approach may indicate common mental processes suggesting a common author.
Inferring Characteristics of Authors
Given the difficulty in pinning down unique aspects of personal style, it is not surprising that attempts to link writing style to personality and other characteristics has been even less successful. The difficulty does seem to be the need, somehow, to take account of the particular contexts in which the utterances occur. For example, even threats may be expressed differently, depending on the type of threat being expressed. People are likely to reveal their traits in quite different ways when trying to bully an associate into keeping away from their girlfriend compared with when they are trying to extort money out of a supermarket chain. Following this line of thought, there has been moderate success in recent studies. For example, people who used more profanities in personal threat letters did have personality profiles that showed a greater desire to control other people. But clearly there is a long way to go before this line of study can be precise enough to be really useful to a police inquiry.
The possibility for distorting the modes of expression in written material, trying to hide the characteristic style of the writer, is another factor that always needs to be considered. But these attempts may reveal more than they hide. It is difficult for unsophisticated writers to emulate sophisticated ones, but people who have fluency and proficiency find it difficult to totally hide that. In one murder case a suspect had kept a careful diary in which her visit to the victim was carefully recorded in the most benign, casual terms. However, the suspect had been so obsessional in the way she kept the diary that it was clear she would not normally have recorded such a passing visit in the way she did. Acareful analysis and count of all the references in the diary and how they were handled demonstrated that the key entry was unusual in some striking ways, in part because the text it contained was so unremarkable. This analysis was drawn on by the prosecution counsel to shape his crossexamination on the diary, but never presented as evidence. The suspect was convicted of the murder.
Predicting Consequences
Language can be analysed at many levels of complexity: the number of particular types of word, the patterns of certain combinations of words, the forms of grammatical structure used, the types of idiom and the themes that are expressed. Even the approach taken to the task of communication may be examined, how carefully planned it is, what prior knowledge it implies, and so on.
It is the higher level, more general forms of analysis of themes and approach that may be especially useful for answering the questions about what the writer is likely to do; how is he or she likely to react to different forms of intervention?
The majority of people who write threatening letters never follow through on their threat. The act of writing is the expression of anger or frustration, malice or spite. Against that backdrop, the task, then, is to detect those letters that reveal a determination to act. Some explorations do suggest that a number of aspects of the themes and approach to the writing of threats can be used as criteria to predict action. It is not appropriate to publish these in a public forum, but it is worth mentioning that they draw upon a careful analysis of the credibility of the threat and the benefits and costs to the writer of carrying out the threat.
In terms of strategies for dealing with threats, police forces therefore need to distinguish carefully the exact nature of the promised threat. Quite different issues and modes of analysis will be appropriate for individualized threats for personal revenge, political threats on ideological grounds and extortion for money, although one type of threat may masquerade as another. Meticulous study of the form of words in which a threat is expressed can be of great value in understanding what is likely to be characteristic of writers’ forms of expression, what they are really trying to achieve, what sort of people they are likely to be, and the probability of their doing anything other than writing such letters.
Conclusion
In conclusion it has to be emphasized that, despite the claims of a few academics, there is still no strong evidence that a person’s writing style can be uniquely pinpointed, measured and used for comparison with contested examples. There has been no legal case in which forensic psycholinguistics has been accepted without question to establish or disprove the identity of a defendant. It seems most unlikely that we will ever establish unique ‘fingerprints’ in written or spoken styles of communication. The way we express ourselves in words is just too flexible to allow that degree of precision. However, although its use in court may be limited, there are many ways in which careful psychological and linguistic analysis of the crime scene of a written document can help investigations.
In all other aspects of forensic psycholinguistics the analyses at present tend to be rather ad hoc, depending on the very particular circumstances of the utterances. Whether there will ever be the possibility of a generic process for answering any of the forensic psycholinguistic questions is still a matter of debate.
