Compliance and surveillance officers are ramping up their efforts to stamp out inappropriate speech - and not just illegal market-manipulating speech - on the part of traders at banks, according to new findings from Fonetic.

Fonetic, a communications surveillance software vendor, recently logged a 20% increase between this quarter and the same quarter last year in the number of banks asking it to give them software to detect racist and sexist slurs. Compliance officers often ask it to set its software to detect such speech and inform them by sending them alerts, and then to ensure that those alerts go to compliance departmetns. Some data drawn from such alerts shows that traders swear and/or use derogatory language eight times more frequently than they express any intention to 'abuse' the markets.

Financial institutions are coming under more and more pressure from regulators to reform their cultures, especially in the UK with its Senior Managers Regime, which aims to make executives at banks personally accountable. Fonetic believes that the 'MeToo' Hollywood propaganda campaign (which has exposed instances of sexual harassment and abuse in show business) has trickled down into financial services, prompting banks to seek out male employees who have done something sexually or racially inappropriate at work and punish them. It also believes, however, that the financial sector is still tarnished by long-standing misogyny and the "old boys' club." Nevertheless, it is of the firm opinion that swearing and sexist remarks are "unacceptable" and that "financial institutions are in no position to brush it [such speech] under the carpet." It believes that there are other "movements" than 'MeToo,' although it does not identify them, and that these movements are prompting compliance officers to redouble their efforts to wipe out any inappropriate behaviour, even behaviour that breaks no laws.

Surveillance software, of course, provides banks with the means to identify politically incorrect talk or, as Fonetic puts it, "the severity of conversations." Natural language processing, Fonetic points out, can now put the things that traders say into context, so even oblique put-downs and insults are detectable. Surveillance software typically works either by detecting inappropriate speech according to a set lexicon of words or by applying machine learning to monitor conversations to detect sinister intent.