Theoretical approaches to speech production There are two main theories of speech production, the Spreading Activation Theory - SAT (Dell, 1986: Dell & O'Seaghdha, 1991) and the Word -Form Encoding by Activation and Verification – WEAVER++ (Levelt et al., 1989: 1999). The SAT theory was created by Dell (1986) then revised by Dell & O'Seaghda (1991). The theory works on a 4-level connectionist model: parallel and dynamic. The semantic level is the meaning of what will be said. The syntactic level is the grammatical structure of the words in the intended utterance. The morphological level is the morphemes (basic meaning units of word forms) in the designed sentence. The phonological level is the basic unit of sound within a sentence. In addition to the main structure of the SAT model, a representation is formed at each level. Pre-planning is more detailed at a semantic level. There are categorical rules at each level, which impose constraints on the categories of elements and the combination of categories. The internal lexicon (dictionary) is considered a constructionist network that includes nodes for concepts, words, morphemes and phonemes. Then, when a node is activated, it sends a message to activate all other nodes connected to it. The subsequent WEAVER++ computational model was proposed by Levelt, Roelofs, and Meeyer (1999) derived from Lefvelt (1989). The model is based on the assumption that there is a feed forward activation network that spreads through the network and does not come back. There are 3 levels in the network: the highest level of nodes represented are lexical ones, the second level are the lemmas which are words abstracted from the mental lexicon and the lowest level are the nodes supporting the morphemes, the basic unit of meaning and phonemes. The network has no inhibitory connections. Speech production follows exactly serial phases. A word error will occur if the level of activity in the node does not match the appropriate node higher up. There are 6 processing steps in WEAVER++ theory 2 more than in SAT theory. First the conceptual preparation phase in which lexical concepts are activated. Phase 2 comes lexical selection where an abstract word or lemma is selected along with its syntactic features. Morphological coding is the basic form of the word derived from the activated lemma. Phase 5 is phonetic coding, where speech sounds are set. Articulation is the final stage that defines how the word is pronounced.