It's hard to state exactly when the specialised role of ‘defensive midfielder' first came into being. In football's early days, the game was focused more purely on goals, as the first international match between England and Scotland, from 1872, testifies. England played 1-2-7; Scotland played 2-2-6. Even the two midfielders were not defensive, per se. England's strategy was to use their superior strength and dribble past the Scots, who went for a passing approach. Thus, one could categorise the England midfielders as "box-to-box" and the Scots as more "deep-lying playmakers", if there's a need for modern terms.
Even as teams made the move to 2-3-5, the defensive midfielder didn't yet come into play; the ‘centre half' position only came into existence when teams started dropping one of their centre forwards into the midfield. A this point, the centre-half was not yet a centre back, but was, in fact, a midfield player. As Jonathan Wilson writes in Inverting The Pyramid, the centre-half was then "the fulcrum of the team, a figure far removed from the dour stopper he would become. He was a multi-skilled all-rounder, defender and attacker, leader and instigator, goal-scorer and destroyer". Yet, as Wilson says, the centre-half would eventually become a dour stopper, the precursor to the specialised defensive midfielder. In fact, the ‘wing halves', the other two members of the midfield three were more defensive, as they were required to pick up the opposition's inside forwards, but even they were expected to contribute going forward, with the only purely defensive players being the two fullbacks.
From there, the development of the defensive midfielder splits off; he developed in different forms in each of Argentina, Brazil, England and continental Europe. What is remarkable, though, is that although there was a much smaller exchange of ideas between countries and continents, nearly every country had a pattern of using three defensively minded players by the first World Cup, in 1930.
Unsurprisingly enough, the characteristics of this extra defensive player related very much to the footballing characteristics of the country he played. In England, the pioneer of the changes was Arsenal manager Herbert Chapman. As Jonathan Wilson referenced, Chapman took the decision to drop his centre-half from midfield into the defence, moving him in line with the two full backs as a means of combatting and exploiting the new offside law. (In the process he created the W-M formation, which would be a staple of British football for over 30 years). Most centre halves in England weren't classy, ball-playing defenders, but rather big and strong physical players, well suited to stop the direct aerial game, meaning the tendency for attacking players was to try and either dribble or batter their way through on goal.
Chapman's centre-half, who had now become the centre back (and will be referred as such hereafter, with centre-half meaning midfield player), was Herbie Roberts. Tom Whittaker, the assistant manager to Chapman, said that Roberts' job was "to intercept all balls down the middle, and either head them or pass them short to a teammate". Thus, Roberts was purely a defensive player, and was expected, and told, to contribute very little in attacking play. While Roberts was not a defensive midfielder, he was one of the earlier examples of a player being given a specialised defensive role.
It was Italy, though, which saw the debut of what was perhaps the first true defensive midfielder. Vittorio Pozzo, who oversaw the metodo system, had been inspired by his time in England, where he greatly admired, and struck up a friendship with Manchester United's centre-half Charlie Roberts. When Pozzo again became the manager of the Italian national team in 1924, he dropped Fulvo Bernadini because he was more of a ball carrier-box to box-than a passer. Before the 1934 World Cup, Pozzo would select Luisito Monti because of his mobility and ability to cover ground. Monti played deeper than the other two wing-halves, and, in the defensive phase, would drop deep to mark the centre forward, before pushing forward to midfield, but always remaining the deepest midfielder.
Yet, up to this point, playing defensive midfielders was part of playing defensive tactics. Herbert Chapman's Arsenal were a counter-attacking side, who were criticized when lesser teams tried to copy their style and failed to do so, producing extremely dull and defensive football. This, of course, is an Arsenal blog, and Arsenal in their current construction aren't a defensive team. While none of the tactics described thus far relate directly to Arsenal today, the development of the defensive midfielder does relate. An even stronger connection can be found when exploring how more attacking teams set up; after all, balance needed to be found, even in the freewheeling days of 2-3-5.
And nowhere can that be better found than in the nations that dominated South American football: Argentina, Uruguay, and Brazil. The former two, the finalists of the first World Cup, were heavily influenced by Italy, with the flow of immigration from there, and in turn the increased movement of South American footballers to Italy, resulting in an exchange of ideas. As a result, both Argentina and Uruguay avoided the development of the third centre-back, using the Italian metodo style. As Rob Sweeney wrote in Issue One of The Blizzard, "the Uruguayans had somehow eschewed Chapman's third-back game and simultaneously demonstrated their defensive robustness, all while upholding their commitment to elaborate football". While Argentinian football was more attacking, and less dedicated to defensive robustness, the Uruguayans struck the perfect balance between attack and defence, enabling them to win the Gold Medal in football in 1924, 1928 and the World Cup in 1930. The Brazilians, winners of only two of the first twenty South American Championships, looked to the south to improve their defensive structure, while maintaining attacking fluidity.
They inherited the Italian metodo, with a slight twist: the Brazilian centre half was, according to Sweeney, "a disciplined holder who was also free to advance with the ball when in possession -a libero, in the most creative sense of the word". Yet, this more attacking centre half left Brazil defensively vulnerable, which is how the more tactically sophisticated Uruguay dominated early South American tournaments.
It was actually a Brazilian, Gentil Cardoso, who claimed to had watched Herbert Chapman develop the W-M at Arsenal, who tried to introduce the thoroughly English shape to the Brazilian game. However, he managed clubs far too small to make any greatly significant impact. The man who introduced the beginning of the third centre back (despite resistance from his centre-half at Flamengo, Fausto dos Santos) was Dori Kurschner, managing Flamengo as well as advising the national team at the 1938 World Cup. Brought up in Hungary, Kurschner came from the Danubian School, and thus, his concept of W-M would've been more attacking than the English style from the start; with the centre-half stationed a little further forward of the two full backs, it was far more like metodo. The defensive openness, though, caused Brazil to concede 10 goals in 5 games, and though they finished third, their run involved a ridiculous 6-5 extra time win over Poland in the first match.
Meanwhile, at Flamengo, Kurschner got the sack after losing the first game in the 1938 Carioca Championship, 2-0 to Vasco da Gama. This had followed finishing 2nd in 1937, despite scoring 83 goals, to eventual title winners' Fluminese's mere 65. Yet, Fluminese conceded only 22 in 22 games, while Flamengo conceded 34, and they ended 3 points apart. Kurschner would be replaced by his assistant, Flavio Costa. Instead of returning to 2-3-5, Costa went for a sort of compromise, what he would call a ‘diagonal'. One of the wing-halves was pulled back, depending on where the attack was coming from, or on what wing the opposing side was stronger. The centre-half Volante - whose name would eventually be given to the defensive midfield position in Brazil - operated as the pivot, and dropped almost into the defensive line when Flamengo were out of possession. To make up for the shortfall in midfield, one of the inside forwards dropped deeper, but retained his creative, attacking nature.
The system foreshadowed what was to come: three defenders; a fourth, very defensive midfielder, and then two more midfielders behind four attackers. Once Volante was shifted slightly deeper, the system evolved into a 4-2-4. With the diagonal, Flamengo would win three titles in a row. The system became more and more popular, growing into Brazil's system of choice for the 1950 World Cup; defeat there, though, would see the fourth defender finally introduced, and would end the era of 2-3-5 based systems in the country.
Part Two: The defensive midfielder in the back four
Inverting The Pyramid, by Jonathan Wilson
The Blizzard, Issue One: Brazil and the Rise of the Back Four (pp 104-122), by Rob Sweeney