Opinion
Crim. No. 82-38.
July 20, 1982.
W. Thomas McGough, Asst. U.S. Atty., Pittsburgh, Pa., for plaintiff.
Robert McClenahan, Pittsburgh, Pa., for defendant.
SUPPLEMENTAL OPINION
I. Introduction
Defendant Russell Clifford was indicted by the Grand Jury in a two count indictment: Count I charged that on or about December 29, 1980, in Saltsburg, Pennsylvania, Defendant Clifford knowingly caused the United States Postal Service to deliver a communication [with hand printed letters in block] form containing threats to injure the persons of Charles Sharon, and others, in violation of 18 U.S.C. § 2 and 876; Count II charged that on or about January 2, 1981, in Saltsburg, Pennsylvania, Defendant Clifford knowingly deposited in an authorized depository for mail matter a communication [with hand printed letters in block form] and addressed to Charles Sharon, to be sent and delivered by the United States Postal Service, which contained threats to injure the person of Charles Sharon and others, in violation of 18 U.S.C. § 2 and 876.
On March 26, 1982, the United States filed a Motion requesting that Defendant Clifford be required to furnish further handwriting exemplars to the Government. On March 2, 1982, Defendant Clifford had previously furnished to the Federal Bureau of Investigation certain hand printed exemplars for comparison with the threatening communications. During the course of the investigation, the United States obtained various pieces of handwritten (cursive) correspondence allegedly written (but not block printed), authored, and signed by Defendant Clifford. However, the hand printed exemplars previously obtained were not suitable for comparison with the handwritten (cursive) correspondence, and Defendant, through his counsel, refused to stipulate that he authored the handwritten (cursive) correspondence. The government's intention in procuring the additionally requested handwritten (cursive) exemplars was to identify Defendant Clifford as the author of the handwritten (cursive) correspondence, and then to compare that known handwritten (cursive) correspondence with the block printed threatening communications, in an attempt to show Defendant's authorship of the threatening communications by means of certain similarities in spelling, abbreviation, syntax and paragraphing between the block printed threatening communications and the cursive handwritten correspondence. This methodology of handwriting comparison was labeled "Forensic Linguistic Analysis".
Argument on that motion was heard April 14, 1982, and the Court took that motion under advisement, pending submission of briefs by the parties. On April 26, 1982, the United States filed a Motion in Limine, seeking a ruling pertaining to the admissibility at trial of certain documentary evidence. These documents were identified as Exhibits B-1 through B-7, consisting of written letters, some with envelopes, which were addressed to various people and officials of Saltsburg, Pennsylvania; Exhibit C consisted of the block printed handwriting exemplars provided by Defendant Clifford on March 2, 1982.
During the course of the April 14, 1982 argument, and after the United States filed its Motion in Limine, it appeared to the Court that the analysis of similarities in spelling, abbreviation, syntax and paragraphing which the Government intended to make at trial pertained to the aforementioned methodology known as "forensic linguistic analysis." Therefore, in order to make an informed decision regarding the propriety of requiring additional handwritten exemplars as requested by the Government, and in order to properly rule on the Motion in Limine, this Court entered an Order on May 17, 1982, requiring a hearing on the use, and present state of the art of the "forensic linguistic method", to determine whether this type of methodology presently was sufficiently reliable and trustworthy to warrant its use in the above-captioned criminal matter.
That hearing was held June 8, 1982, and expert testimony was furnished by several agents, employees and experts of the Federal Bureau of Investigation. Following that hearing, and after consideration of the briefs previously submitted, as well as the testimony presented at that hearing, this Court orally ruled from the bench that the Motion for Production of Further Handwriting Exemplars was denied for the reason that said new methodology, namely, "Forensic Linguistic Analysis" had not been advanced to the point wherein the state of said art was sufficiently trustworthy and reliable to be used in a criminal case. The Court further ruled that the Government could not introduce at trial the documentary evidence offered in its Motion in Limine. The purpose of this Supplemental Opinion is to clarify, explain, and delineate the basis of the Court's June 8, 1982 ruling.
II. The Forensic Linguistic Method
At the June 8, 1982 hearing, the Court received the testimony of Dr. Murray S. Miron, a professor at Syracuse University who renders psycholinguistic services to the Federal Bureau of Investigation; Penelope Pickett, an aural analyst with the Federal Bureau of Investigation Laboratory, and the author of the FBI report denoted as Court Exhibit 1-A; and Special Agent Ronald M. Furgerson, the unit chief in charge of the technical evaluation unit of the document section of the Federal Bureau of Investigation Laboratory.
Dr. Murray S. Miron testified that he is under contract with the Federal Bureau of Investigation to provide psycholinguistic services in cases involving disputed or unknown authorship. These analyses include psychiatric characteristics of the author, demographic characteristics of the author (age, economic, background, residence) and linguistic analysis based on syntax, punctuation, and spelling. To accomplish this, Dr. Miron has developed computer programs which compare vocabulary and stylistic variants, e.g., length of sentence, number of periods, exclamation marks, dependent clauses. Additionally, each vocabulary item is assessed with respect to its frequence of occurrence in the language. Certain misspellings and certain abbreviations would be weighted heavily in the comparison, while others would be lightly weighted or not weighted at all. Transcript, June 8, 1982, at 8 10.
Dr. Miron uses standard counts of present American English which list the frequency of occurrence of more than one million words of text taken from fiction, novels, newspapers, and other printed sources. In determining whether a certain misspelling is a significant factor, reference is made to this sample, and if that particular misspelling does not appear, such misspelling is considered to be significant, and the word is known as a signature word, that is, a distinctive word that is idiosyncratic or peculiar to a particular writer. Transcript, June 8, 1982, at 10-13.
Dr. Miron concurred with the identification of the significant features of the known and unknown samples that were made by the Federal Bureau of Investigation expert in Court Exhibit 1-A because the misspellings were unusual. Tr., June 8, 1982, at 20.
Dr. Miron also stated that the application of these methods as evidentiary proof of authorship was premature, and the author of the threatening communications could not be identified to the exclusion of any other possible author. Part of the problem in determining authorship is that the samples involved were short segments, and since language shows high degree of variability, larger samples were required to insure that there is a representative sample. Dr. Miron also stated that he did not consider the method to be suitable for testimonial purposes to establish certainty of authorship, although he stated expert testimony could be used to point out to a jury how unusual certain similarities or dissimilarities are. The Court concluded from this testimony that the "Forensic Linguistic Analysis" methodology was insufficiently advanced as a trustworthy art to warrant its submission with or without an expert, to a jury.
III. The Government Contentions
According to the Brief of the Government, a comparison of the anonymous threatening letters with known samples of Defendant Clifford's handwriting reveals certain striking similarities. The Government points to misspellings of "explosives" as "explodsives" "figure" as "figuar"; the inconsistent use of either circles or dots for the letter "i"; the lack of indentation or space between paragraphs in all documents; and the use of certain abbreviations in all documents — "Charles" as "Chas.", "A.M." as "AM", "Street" as "ST", "Pennsylvania" as "PA", and the days of the week as "Tues.", "Wed.", and "Thur.".
It is important to note that the additionally requested handwritten exemplars were to be the basis of the Government's contention that Defendant Clifford in fact authored the handwritten correspondence which was sent to various Saltsburg officials, and designated as Exhibits B-1 through B-7 in the Motion in Limine. It further is to be noted that the purpose of the comparison of the documents is not to find similarities in the handwriting itself, as is done in traditional handwriting analysis, but is to find similarities in word choice, misspellings, paragraphing, style, and abbreviations. Finally, it must be noted that even assuming the Government can prove that the handwritten letters were physically written by Defendant Clifford, it is by no means certain that he actually authored them, that is, the letters could have been dictated to him by someone else, or they could have been copied from some other source.
From pointing out similarities of this nature, the Government seeks to prove circumstantially that Defendant Clifford authored the threatening communications mentioned in the indictment. The Government has specified that it does not intend to introduce at trial any expert opinion or testimony on the likelihood that Defendant Clifford authored the anonymous letters. However, Dr. Miron testified categorically that under no circumstances could a jury comprehend this evidence without the use of an expert and even then there would not be the required certainty that would lead a jury to believe the Defendant was the author beyond a reasonable doubt.
At the April 14, 1982 argument, the Court inquired whether the Government had submitted the known writing to the Federal Bureau of Investigation Laboratory. The Government responded affirmatively, and produced at the Court's request a copy of that report. That report, Court Exhibit 1-A, indicated that a forensic linguistic examination was conducted on the documents submitted. The relevant portions of that report read as follows:
. . . The results indicate a strong likelihood that RUSSELL CLIFFORD is the author of the questioned documents. This conclusion is based on (1) the identical "peculiar" misspelling, "figuar," and the identical misspelling, "Explodsives," in both sets of documents, (2) the presence of relatively long sentences in both sets of documents, (3) the appeal in each set of documents to the receiver's presumed aversion to adverse publicity, and (4) the similarity in paragraphing format. It is noted that "Thur." appears in Qc5, while "Thursday" is the consistent misspelling in K11. Perhaps it could be determined from other known writing samples whether MR. CLIFFORD draws this spelling distinction between the abbreviation and the full word, or even if he sometimes reverses letters in the abbreviation.
You are reminded that since application of the linguistic method is not considered a positive means of identification, results of such examinations are provided for investigative assistance only, and are not intended for testimony.
(Emphasis supplied).
IV. Discussion
A trial judge is vested with discretion in determining relevancy and admissibility of evidence under Federal Rule of Evidence 403. Hamling v. United States, 418 U.S. 87, 94 S.Ct. 2887, 41 L.Ed.2d 590 (1974). Under Rule 403, a trial judge may exclude relevant evidence when its probative value is "substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury." Because Rule 403 gives the trial judge the discretion to exclude relevant but misleading evidence, the ruling of the trial judge will be reversed only if an abuse of that discretion can be shown. United States v. Sindona, 636 F.2d 792, 800 (2d Cir. 1980), cert. denied, 451 U.S. 912, 101 S.Ct. 1984, 68 L.Ed.2d 302 United States v. Aulet, 618 F.2d 182 (2d Cir. 1980). A jury may not convict a criminal defendant on the basis of suspicion, surmise or conjecture, since the burden of proof beyond a reasonable doubt would not be met in that instance. United States v. DuBois, 645 F.2d 642 (8th Cir. 1981); United States v. Heithaus, 391 F.2d 810 (3d Cir. 1968). While the proposed evidence which is offered to show the authorship of the threatening letters in the instant case is concededly relevant to the offenses charged in the indictment, this does not guarantee its admissibility, and this Court holds that such evidence cannot be admissible since the likelihood that the jury will be seriously mislead by such evidence outweighs the probative value of the proffered evidence, inasmuch as said type of evidence i.e., based on a "Forensic Linguistic Analysis" is not reliable and trustworthy at this stage of the development of said art.
As previously stated, the United States Attorney seeks to present to the jury certain similarities and dissimilarities in syntax, spelling and paragraphing styles between the threatening communications which are the subject of the indictment and correspondence which is purportedly authored by Defendant Clifford. The Government has stated its intent not to use expert testimony in pointing out these similarities, and has also stated that it does not intend to introduce at trial Court Exhibit 1-A, the expert report of the Federal Bureau of Investigation Laboratory concerning the forensic linguistic analysis. The proposed use of this type of evidence under these circumstances is clearly unreliable and untrustworthy. See testimony of Dr. Miron.
After considering the evidence presented at the June 8, 1982 hearing, this Court finds as a fact that "forensic linguistic analysis" consists of an in depth evaluation of linguistic characteristics of text, including grammar, syntax, spelling, vocabulary and phraseology, which is accomplished through a comparison of textual material of known and unknown authorship, in an attempt to disclose idiosyncracies peculiar to authorship in an attempt to determine whether the authors could be identical. See Government Exhibit 4, Job Description of Aural Analysis GS-11. See also Pickett testimony at 73-74; Miron testimony at 12-15.
This Court is of the opinion that to allow the United States Attorney to present the proffered evidence to the jury without the aid of an expert would be, in effect, to allow a non-expert to use forensic linguistic analysis and ultimately testify on matters solely within the realm of expert testimony. This the Court will not allow as the jury would be seriously mislead. Dr. Miron stressed in his testimony that without the aid of an expert the jury could easily misinterpret the significance and the weight to be applied to either similarities or dissimilarities in the textual materials. Transcript, June 8, 1982, at 41-42.
The possibility that the jury would be mislead by a study of similarities and dissimilarities with or without the aid of an expert is very strong. For example, it is obvious to the Court that the addresses on the envelopes of the threatening communications read "Chestnut St" while the exemplars provided by Defendant Clifford read "Chesnu St" an obvious dissimilarity which is not mentioned in the Federal Bureau of Investigation Report. When presented with this situation, Dr. Miron testified in his opinion that the difference in spelling would be evidence of one area of dissimilarity, and if that were the only evidence he would conclude that the Defendant was not the author of the threatening communications. Transcript, June 8, 1982, at 34. However, Dr. Miron went on to conclude that in his professional judgment he would not count it as a significant difference, that is, it would be a difference but one that he would not weigh heavily. Transcript, June 8, 1982, at 37-39.
The reason Dr. Miron would not consider it to be a significant misspelling is "because neither word [Chestnut or Chesnu] would occur in the [frequency] count with respect to this. I would not try to make a determination of the frequency of this word because I would not expect the name of a street to occur. They did not count one million words in Saltsburg, Pennsylvania". Two other misspellings of explosives and figure would however, be heavily weighted by Dr. Miron as similarities which would lead him to the probable conclusion that the communications were written by the same author.
It is readily apparent to this Court that it is not the similarities or dissimilarities which are controlling, but it is the weight to be given to them which is the crucial factor in performing forensic linguistic analysis. Obviously, the determination of the weight and significance is something which is beyond the realm of an ordinary juror. To place the question before the jury with or without the aid of an expert would be seriously misleading and would lead to the dangerous situation wherein the jury verdict would be based on surmise and speculation, and not on proof beyond a reasonable doubt which is the constitutional standard required in all criminal cases.
This Court has also considered the possibility of permitting expert testimony on the similarities and dissimilarities in the writing, and concludes that the use of expert testimony would not materially enhance the reliability and trustworthiness of this type of untried evidence. After an extensive review of the law applicable to the use of scientific and expert testimony in criminal cases, this Court has found only one case wherein a similar fact situation was presented to the trial court. In United States v. Hearst, 412 F. Supp. 893 (N.D.Cal. 1976), the defense sought to admit the testimony of Dr. Singer, a psycholinguist, who would have testified that her expertise enabled her to "conclude from stylistic comparison of known writings and utterances of the defendant with certain writings and tape recordings of the defendant's voice offered into evidence by the Government that these latter writings or utterances could not have been authored by the defendant." 412 F. Supp. at 894.
The Hearst court ruled against the admissibility of such testimony, not because the expert was not qualified in the field of psycholinguistics, but because the state of the art of psycholinguistics was such that it had not achieved such general acceptance among psychological and scientific authorities to justify its admission. To allow its admission would therefore have created an unjustifiable "aura of special reliability and trustworthiness". 412 F. Supp. at 895. This exclusion by the trial court was held proper on appeal. See United States v. Hearst, 563 F.2d 1331, 1349 (9th Cir. 1977) (rehearing denied).
This Court is not persuaded that in the instant case the trustworthiness and reliability of the forensic linguistic methods has been established or that the method has gained general acceptance in the scientific community since the Hearst case was decided.
The Federal Bureau of Investigation itself considers the results of the linguistic method to be inappropriate for testimonial purposes. See Court Exhibit 1-A, Federal Bureau of Investigation Lab Report ("You are reminded that since application of the linguistic method is not considered a positive means of identification, results of such examinations are provided for investigative assistance only, and are not intended for testimony"). (underlining supplied). It is also important to note that Penelope Pickett, the Federal Bureau of Investigation aural analyst who authored Court Exhibit 1-A, is not certified by the Federal Bureau of Investigation as being an expert for trial purposes. Special Agent Ronald Furgerson, the unit chief in charge of the technical evaluation unit of the document section, testified that because of the laboratory policy relating to linguistic analysis, there has been no need to qualify her as an expert since the basis for the services is for investigative assistance, without regard to what may happen in the judicial system past the investigation stage. Transcript, June 8, 1982, at 53-54. Special Agent Furgerson also stated that because of the unknown variables in the field, the laboratory uses forensic linguistics for investigative guidance only and not for positive identification. Transcript, June 8, 1982, at 66. Special Agent Furgerson also testified that he would be uncomfortable if the linguistic method were used for positive identification. Transcript, June 8, 1982, at 69-70. Ms. Pickett, who prepared the FBI report, Court Exhibit 1-A, testified that she could not make a positive identification of the Defendant as the author of the threatening correspondence, only that there was a "strong likelihood" that the Defendant was the author of the threatening communications. Transcript, June 8, 1982, at 75-76.
Although Federal Rule of Evidence 702 permits the admission of expert testimony where it will aid the trier of fact, the right of a criminal defendant to a fair trial is a countervailing restraint on the admissibility of the evidence under Rule 702. United States v. Green, 548 F.2d 1261, 1268 (6th Cir. 1977).
"The fate of a defendant in a criminal prosecution should not hang on his ability to successfully rebut scientific evidence which bears an `aura of special reliability and trustworthiness, although in reality the witness is testifying on the basis of an unproved hypothesis in an isolated experiment which has yet to gain general acceptance in its field." United States v. Brown, 557 F.2d 541, 556 (6th Cir. 1977) (inadmissibility of results of ion microprobic analysis of human hair).
Forensic linguistic analysis has not been shown to be a reliable method of ascertaining authorship, as evidenced by the fact that the Court is unable to find any instances wherein testimony pertaining to the linguistic method has been permitted, and further evidenced by the fact that the Federal Bureau of Investigation is unable to use the results of the linguistic method as positive identification.
Forensic linguistic analysis is unlike fingerprint identification, ballistics identification or even voice spectrograph where the subject of that analysis is considered by the scientific community to be a unique characteristic. Further, once a similarity or dissimilarity is found, the forensic linguistic analyst uses his subjective judgment and expertise to assign a weight of significance to each particular difference, thus giving rise to an element of speculation which is not present in the other mentioned tests. The potential for the jury to be mislead by the use of a linguistic analysis is ever present.
Finally, this Court is cognizant of another danger present in the use of the linguistic method, namely the problem of determining authorship in a questioned document. The questioned document or sample could have been copied from another source, dictated by another person, or deliberately disguised in an attempt to imitate distinctive styles which would implicate a third person. See Testimony of Furgerson, Transcript, June 8, 1982, at 57.
In summary, this Court has exercised its discretion and holds that linguistic analysis is not admissible in the instant case, with or without the use of expert testimony. It is inadmissible as expert testimony because the method has not been shown to be trustworthy, reliable, accurate, or conforming to a generally accepted scientific theory. The probative value of the testimony is outweighed by the potential that the jury will be mislead and defendant's right to a fair trial is further jeopardized by the undue weight such testimony would carry with the jurors, which would result in possible mistake or confusion.
The evidence is also inadmissible without the aid of an expert. Dr. Miron concedes that such testimony could not be evaluated properly by the jury without the aid of an expert. The evidence itself is inherently untrustworthy, and this inherent defect cannot be cured by presenting the testimony without the aid of an expert.
The Court therefore denies the request of the Government for additional handwriting exemplars; and after exercising its discretion, the Court holds that the Government's Motion in Limine is also denied, and the Government will not be permitted to introduce known exemplars of the Defendant's writings for the purposes of stylistic comparisons with the anonymous communications.