Although cinema is discussed primarily in terms of sight and sound, the latter element is usually considered secondary: a mere embellishment, or an adjunct to a film's visual identity.
If sound design and music only occasionally command our attention and scrutiny, we are even less likely to think about voice. Do we take it for granted because our relationship with voice pre-dates by almost a year any optical engagement with the world? Has voice been with us for so long that we can fail even to notice it?
But if it is through voice (as well as touch and smell, senses that cinema has yet to replicate adequately) that we forge our primal bonds, no wonder the sound of a particular actor can trigger emotional associations that render us powerless.
Kubrick's Barry Lyndon is one of my favourite films, and a work to which I return regularly and willingly. But I wonder what percentage of my positive response to that picture can be attributed to the presence of Michael Hordern as the unseen narrator. You see, Hordern provided the voices for the BBC's Paddington Bear ; I was four years old when that animated series -- made in bite-sized, five-minute episodes screened at teatime -- began in 1975 (the same year, coincidentally, that Barry Lyndon was released).
Consequently, Hordern's voice joined the aural collage of my childhood, and I have no doubt now that something in me responded to his gently rumbling timbre when I first saw Kubrick's picture in my early twenties. And yet it only occurred to me when I started writing this post.
Voice is vital to our relationship with character and actor, even when we cannot understand what is being said (Benicio del Toro in The Usual Suspects), or when the voice does not belong to the on-screen body (James Earl Jones as Darth Vader in the Star Wars films). In his astute appreciation of James Stewart, published in the London Review of Books  in 2002, David Bromwich insists that the contours of Stewart's voice are crucial to our understanding of that cherished star:
One thing a casual viewer learns to love, if he is going to like Stewart at all, is a kind of stammer that trips in naturally and convincingly -- a signature touch he seldom allowed to pass into self-parody until his late fifties. An anomaly almost as emphatic is the frequent decision to speak in a soft voice, always with perfect clarity and conveying a range of available senses for words. Stewart does this often in intimate scenes with women, but not only with them, and it shows the passage from theatre to a broader naturalist domain of feeling that the movies uniquely made possible. Even now, when fewer actors bring the wrong kind of theatricality from stage to screen, the freedom to modulate a speaking voice downward is rarely grasped; and if you listen to the better-known stars of the 1930s and 1940s, only a select company of them appear to have glimpsed and taken the opportunity: Ida Lupino, Barbara Stanwyck, Henry Fonda -- were there many more? Even within that group, Stewart is exceptionally resourceful. His voice can be put in the service of feelings as they bubble slowly from confusion to clarity. Or it can be used to signal the intimation of half-thoughts, shadowy promptings of a kind that only a first-rate writer may catch in words.
The voice has attracted more attention with the rise of animation in the past two decades, and in particular the celebrity voice-casting that has predominated ever since Robin Williams, as the Genie in Disney's Aladdin, turned a simple shift in the dubbing studio into a gutsy, full-blown performance. (His was one of the first vocal turns to dictate the direction of the animation, something that now happens routinely.)
But while a few hours at the microphone in the service of an animated film has become part of the career plan of any Hollywood star, I still don't know if we appreciate the extent to which voice can enhance and even define a movie. Tom Hanks won his brace of Best Actor Oscars for Philadelphia and Forrest Gump. Yet I regard his finest work, and the film in which he seems most emotionally present despite his physical absence, as being his portrayal of Woody the cowboy doll in the Toy Story series. In the tenderness of his line readings, he isn't lip-syncing, or layering his voice over the film; voice and image become indivisible. He is Woody.
There is a similar synchronicity in the vocal performance by the comedian Steve Carell in the upcoming computer-animated film Despicable Me , which I saw recently. (It opens in the UK in October.) To play the super-villain Gru, who is trying to steal the moon in order to trump his nearest rival in the business of evil, Carell has concocted a vocal tangle that suggests a history of movie wickedness. There are some German consonants in there, and a peppering of Russian, but most of all it's that all-purpose Hollywood accent known as "Foreign".
Carell suffuses this with tremors of vulnerability -- like Dr Evil in the Austin Powers movies, Gru is essentially an anxious figure, middle management in essence, who would really just like to be considered good at what he does. The film's 3-D animation will be one of its major attractions, but it would feel flat without Carell.
For more on this subject than can reasonably be covered in any blog, do grab a copy of Michel Chion's groundbreaking book The Voice in Cinema , in which the author also considers the phenomenon of the disembodied voice (HAL in 2001: a Space Odyssey, "Mother" in Psycho). For the best results, I recommend reading it aloud.
Ryan Gilbey is the New Statesman's film critic . He blogs at Cultural Capital every Tuesday.