The process of communication is not one involving the simple model 'sender/ message/ receiver', but one involving production, circulation, consumption, and reproduction. It is a complex 'structure in dominance' where each level is relatively autonomous. There is a passage of forms [as well as content].  Meanings are produced and organised as codes in a syntagmatic discourse [roughly, one which unfolds over time and explains meaning by a system of differences and unfoldings]. Symbolic vehicles are produced and circulated: they must then be translated back, read, and articulated into practice. There can be, of course, no guarantees [of the success of this] process. 

Once events are coded, they are subject to discursive rules, which dominate communication (although the event still exists -- [Hall is unwilling to go the whole way with then-fashionable views that 'nothing exists outside the text' -- see my comment on the later works by Barthes here). The message form, therefore, faces two ways  - it is a mere surface form of a communication system and part of a general communicative system [that is, it is no use analysing single 'messages' as self-contained units of meaning]. 

Let us consider the production of meanings first, to illustrate this point. It has itself a  'discursive aspect' [that is, it is no use seeing messages as just the personal and subjective expressions of individual producers either]. It involves institutional knowledge, professional ideologies, images of the audience, and feedback, with links to consumption as well. These stages are only moments in a [communicative and cultural] totality (page 130). 

When a message is produced, it must be meaningful. Production thus also implies decoding practices -- there is no simple behavioural process involved in the transference of meaning [as in an earlier tradition of media research -- 'effects analysis'].  'Effects' occur within structures of understanding. Decoding structures need not be the same as encoding ones, though, and it is this that has led to an exciting new phase in audience research [at Halls' own Centre, no doubt] to replace the  'low- flying behaviourism' [of effects analysis] ( page 131). For example, it is not violence itself but  'messages about violence' which one finds on television [this is a criticism levelled against work like Belson's -- which see]. 

The signs on television are both visual and aural. They are iconic in the sense that Peirce uses that term [fashionable hat-doffing rather than a serious attempt to use Peirce]. They are not transparent representations of the real: 'The dog in the film can bark but it cannot bite' (page 131) -- [a reference to a rather obscure poetic aside by Barthes]. Meaning arises not from simple representation, but from a process of  'the articulation of language on real relations and conditions'. In other words, there is always a code, there is no  'degree zero' [or actual starting point before any coding] of writing. Realism, for example, is an effect of articulation [and not an innocent matter of just recording 'reality' -- see file]. Some codes are naturalised, and this conceals [these coding] practices. 

This assumes a convergence between encoding and decodings, [which can arise in the following circumstances...] Iconic signs are still arbitrary, but are widely understood nonetheless -- but only because  'the visual codes of perception are very widely distributed', more so than with linguistic codes (page 132). This potential for greater conflict in language arises because language is the site of multiaccentuality and class struggle, according to Volosinov [not the case with visual signs, apparently then -- maybe these have always been dominated by ruling groups?] 

The notions of  'denotation' and  'connotation' [big in early Barthes but now creaking a bit theoretically] represent only an analytic difference -- this is only worth keeping because it helps us identify the least fixed aspects of communication, and [maybe eventually] to exploit and transform fluidity [in the right direction -- to assist struggle]. 

The connotational, or contextual level, is of greater ideological significance, as Barthes shows with his work on fashion and advertising -- these forms of communication referred to [connoted] whole maps of [ideological] meaning. Polysemy [the many potential meanings implicit in the sign] is not the same as pluralism though -- some codes are genuinely imposed [by a  'society/culture'! -- Page 134]. There is a dominant cultural order, although it is not uncontested, and this produces preferred readings or dominant readings, an effect of the structure-in-dominance – but we should not see meanings as simply determined. Dominant meanings have the political and ideological ordering imprinted on them and have become institutionalised. [Hall notes a critique of this view offered by one of his students -- O'Shea -- in an unpublished CCCS paper, which I have still not read!]. 

Any differences [at the decoding end] are seen as 'misunderstandings', and are referred to those dominant readings [under the guise of being errors]: in fact, 'rules of competence and use, of logics-in-use... [are not neutral matters of ‘effective communication’ but] ...actually seek to enforce or pre-fer [sic] one semantic domain over another'. Again, none of this is one-sided or automatic: it requires work to win legitimacy  (page 135).[NB this strange habit of splitting normal words like 'prefer' with a dash arises rather a lot in Hall's work -- perhaps we are supposed to spot some special implication, that this preferrring happens before anything else or takes precedence, as in 'pre-determined'?]

Televisual practice  'rearranges, delimits and prescribes' the relations between signs. But there is always a problem to secure dominant readings. However, communication is never transparent but it is  'systematically distorted' [note 13 says this term refers to Habermas, an unusual acknowledgement of an otherwise largely ignored writer -- page 295]. This is not a matter of selective perception, which is a pluralist way of looking at things. Perceptions are [socially, politically, culturally] patterned. 

There is some reciprocity between encoding and decodings. A correspondence is 'articulated' [eg in securing 'effective communication' communicators try to guess how the audience will decode, as we saw -- any educator must be aware of doing this, of course -- and see comment]. Different combinations are possible -- there is a reference to Parkin on class structure and sets of values, but Parkin's work is not discursive enough, says Hall, and nor is Morley's [these were both early influences on and versions of the interest in 'decoding', and much used in earlier papers on this topic]. Nevertheless there are [still] three possible sources of decoding: 

1. First, dominant hegemonic. This takes place within the dominant code -- these decodings are specific variants with in a dominant code [now turned a  'metacode' page 136]. The specifics include the professional codes of broadcasters, which operate within a hegemonic structure and support it by offering an apparent neutrality. Professional broadcasters share the same class background as ruling groups, however (page 137). Of course, some conflicts take place between them. 
2. Secondly, there is a negotiated code -- a mixture of adaptive and oppositional elements, especially at the situational level and with particular logics -- for example, different attitudes to the issue of whether trade unions have too much control [in Britain -- a big divisive issue of the 1970s, the subject of many opinion polls, and one of the specific items used by Morley to gather audience reactions, I recall]. These are the most common source of  'misunderstandings'. 
3. Thirdly there are globally contrary decodings. These detotalize a message, and retotalize it in an alternative framework: thus a term like  'national interest' reappears as  'class interest'. There is a whole oppositional code. This offers a basis for a  'politics of signification', a  'struggle in discourse' page 138. [Compare this with the work of Pecheux, also the subject of a Working Paper by a CCCS student -- see comment].

This article seems to me to offer the usual contradictions and hedged  bets. Hall tries to reconcile both the apparent universality of discourse and the need to see it as an 'unstable equilibrium', which would permit some possibility of struggle. As usual, a 'mapping' exercise is taking place to locate and privilege Hall's view in and against those of rival perspectives. I am not at all sure what the message actually is here, as a result. What should theorists or activists actually DO as a result of all this analysis?

Things are not helped by the almost complete lack of any actual concrete examples. Instead  we have an abstract medal of the three obvious possibilities [for, against, mixed], still based on the introductory work of Parkin and Morley, despite some theoretical embarrassment at their non-discursive/marxist origins. Where there is an example of a specific variation, it is the old and familiar one of professional communicators, and, again, this arises largely from a theoretical commitment to  'fit in the press' rather than from any study of actual broadcasters.