👋 Hi, I’m Melissa and welcome to my biweekly Note, Language Processor. Every other Tuesday, I dig deep into language & behavior, the limits of technologies, and the connection between what people say and do. Once a month, Signal Lab takes over and reports on these topics for startups and early-stage investors.
Member, Board of Parole: So as you know, [REDACTED], we begin these hearings by, uh, asking, uh, asking you to, to give your board your version of the events, uh, surrounding, um, the crime for which you’re incarcerated. So you may begin.
Inmate: Alright. Um, so I’ll just the – begin the night of, uh, the incident.
Years ago, I designed studies with several Departments of Correction on one of the most difficult and under-theorized problems in society: recidivism on parole.1 These projects were the proof of concept of Behavioral Linguistic Intelligence.
Parole and Recidivism
Parole is a conditional early release from prison before an inmate’s maximum sentence has been served. On or after their official parole consideration date, an inmate meets with the parole board to determine whether an early release is appropriate. If the board decides against release, the inmate will remain in prison, continuing to meet with the parole board at intervals unless or until s/he is found suitable and a firm release date set.
From an institutional perspective, keeping inmates in prison is not always tenable — prison over-crowding is already a problem. In the United States, since 1984, the prison population has grown from 500,000 to 1.8 million, making it the largest prison system in the world.2
Prison over-crowding and record-high rates of recidivism (i.e., failure on parole) go hand in hand, with parole viewed as a solution to the former, stymied by the latter.3 Many mandated prison population reductions come in the form of early release on parole for inmates who have not met all formal requirements, exacerbating this circular problem and putting pressure on the parole assessment process.4
How Parole Determinations are Made
The remit of the parole board is to assess an inmate’s risk of recidivism and make a release decision. This occurs formally during a parole hearing, an inmate’s only opportunity to articulate his case for early release. For each inmate, this event occurs as infrequently as once every fifteen years,5 depending on prior performances. The stakes are particularly high among prisoners serving indeterminate life sentences (i.e., life with the possibility of parole). Within sentencing guidelines and other constraints, parole board members hold the power to set a firm parole date, deny, or defer parole indefinitely.6 This life-altering judgment is made by two to four parole board members reviewing something like eight cases a day, allocating 20 to 200 minutes to each.
Each state operates under independent legal precedents and has the authority to emphasize a different set of factors toward parole release decisions. However, some inmate qualities are evaluated across states, including: attitudes to the crime and criminal past, future plans, assumption of responsibility, and inconsistency.
The board exercises comprehensive release authority, drawing on an overwhelming quantity of static assessment materials and actuarial decision factors available to them. To arrive at a decision, parole board members may hear opinions from the District Attorney on the original case, victims and representatives, prison psychiatrists and officials, and inmate supporters.
The board relies on three decision aids: assessment instruments, the inmate’s (criminal and prison) record, and the inmate’s interview performance as judged by the parole board members. Each of these decision aids are infamously flawed and unreliable. Prediction instruments rarely meet (much less exceed) the rate of accuracy recorded in a century-old rudimentary instrument (Burgess, 1928), and psychiatric evaluations are accurate in no more than one in three predictions of future violence. Unfortunately, attempts at objectivity have led to other biases and injustices, as demonstrated in COMPAS, a self-reporting psychometric instrument, one of the most widely used actuarial risk assessments across Departments of Corrections for release decisions.
The Inmate’s Hearing Performance
The parole hearing is the inmate’s only opportunity to speak for himself on the topics of criminal history, the criminal offense, and plans for reintegration into society. Consequently, these statements are richer sources of insight than everyday language, containing not just linguistic data on one moment in time but a focal perspective on personal life history.
The inmate’s performance in the hearing has long been central to parole members’ decision. And yet, the call for evidential basis and critical analysis on an offender’s actual language has spanned decades, as a “type of behavior often overlooked, or underused” though it offers “a wealth of information” (Smith & Shuy, 2002, p. 17).7
If we are to control the increasing expressions of violence which threaten our society, it is imperative that we seek every technology at our disposal to understand the nature and character of those who would use violence as their weapon. What better source could we hope for in our understanding of such individuals than their own words and the content of their communications?
Moreover, many key parole release criteria are intuitive fits for quantitative linguistic analysis. To fill a parole suitability criteria gap left after a 2008 Supreme Court ruling,8 the California Code of Regulations states that past and present attitude to the crime is key (Title 15 § 24029). The strongest emphasis is placed on an inmate’s ‘insight’ into the commitment offense. Though this criterion is not formally measured, a lack of insight is said to be demonstrated by ‘inconsistent statements’ and ‘minimizing culpability’ (Weisberg, Mukamal, & Segall, 2011). In Oregon, board members are instructed to attune to the inmate’s maturity, stability, and demonstrated responsibility. These are suggested to be expressed both in the content and structure of statements (Oregon Board of Parole and Post-Prison Supervision [BOPPPS], 2009). New York parole board members are encouraged to focus not on paralinguistic features during the interview (which can be manipulated), but rather on what the prisoner is saying as a clearer indication of his capacity for reintegration or risk to society.
Further guidance on any of these criteria was not provided by state parole boards, neither were established or objective processes or tools.
The link between language and parole suitability was assumed, but not elucidated.
Inmate: No, I’ve, I’ve come to believe, uh, that in a midst of chaos there is no order. So, you know, it – when you’re in the middle of doing an act, it’s really hard to stop it in the middle of it and go, oh, wait a minute, hold it. This ain’t right, you know, when your life is so messed up in the beginning. You know? I don’t know if I’m making sense.
Parole board members are ultimately responsible to decide whether or not to release the inmate, based on these criteria. They are instructed to use their judgment and experience to draw conclusions on these points from inmate statements. However, while parole board members may have some training or experience in criminal justice,10 they have none in linguistics or communication, and these are linguistic criteria. The parole hearing exists in order to evaluate the inmate’s statements regarding his or her past crime and future plans.
With the aid of linguistic analysis, parole board members would experience less bias, more objectivity, and greater accuracy in release decisions.
Impression-based judgments in release determinations have gone largely unquestioned because most parole boards are not required to give any substantial account for each parole suitability decision. The “enigmatic criteria for parole, and the broad discretion exercised by parole boards in making release decisions” has been criticized for decades.
In answer to these shortcomings, no assessment instrument explored the linguistic content of inmate statements as a quantitative indication of risk for parole release.
Until….
See you in two weeks.
1 Parolees are twice as likely to return to prison than offenders released at the completion of their term (Weisberg, Mukamal, & Segall, 2011, p. 3). More than 11,000 criminology majors graduate every year, but parole remains one of the most difficult and under-researched areas in criminal justice.
2 Another 700,000 inmates are in local jails and detention centers (The Sentencing Project, 2017). And in that same time, the number of people serving life sentences has quadrupled.
3 California, for instance, saw rates of success on parole drop below 20% (Travis & Lawrence, 2002), leading to a level of prison overcrowding justifying gubernatorial emergency proclamations (Schwarzenegger’s 2006 Prison Overcrowding Emergency Proclamation which did not resolve the issue) and ultimately US Supreme Court intervention in Brown v. Plata (2011), mandating that California reduce its prison population by 40,000 inmates within 5 years, citing prisoners’ constitutional protection from cruel and unusual punishment. For a brief history of California parole particularly, see Lynch, 1998.
4 In 2012, 2047 parole hearings were conducted in Connecticut with a release rate of 75% (Connecticut Board of Pardons and Parole [BOPP], 2015).
5 An amendment made to California Penal Code § 3041(b) by Marsy’s Law (California Victims’ Bill of Rights 2008), allowing a 3, 5, 7, 10, or 15 year denial – an extension from the pre-existing maximum of 5 years. Other states hold hearings as frequently as every year, or as infrequently as every 15, depending on state law and readiness demonstrated in prior performance.
6 As of January 2015, forty-seven out of fifty state legislatures had given parole boards the right to grant “evidence-based” parole releases.
7 In criminological studies, the collection and analysis of prisoner narratives can be aimless or purely anecdotal, while linguistic studies do not offer much complexity beyond descriptive analysis (Copes, 2014; O’Connor, 2000). Both are more often used to give a voice to the voiceless than to make analytical judgments.
8 In the matter of Lawrence, removing the commitment offense as a justification for parole denial.
9 Applicable for inmates convicted after 1978.
10 Members of parole boards are commonly appointed (which may have political implications) and usually bring some past related experience in criminal justice to the table (e.g., public defenders, superintendents of youth correctional facilities, parole or probation officers, correctional consultants, criminal justice program administrators).