Alternative Explanations

Above, experimental evidence for the FUL model was outlined. Its theoretical assumptions were confirmed in very different experimental settings, with behavioural as well as electrophysiological measurements, in consonants and vowels, word-initially, -medially and -finally, across and within languages.

However, the FUL model also has its opponents. By the extension of the binary logic of match versus mismatch into a ternary matching logic of match, mismatch and nomismatch, the FUL model gains flexibility in dealing with surface variation. At the same time it enlarges the cohort of possible word candidates.

Proponents of full specification of items in the mental lexicon (e.g. Johnson, 2004) criticise the FUL model for not punishing a nomismatch. They claim that this way the model is no longer able to distinguish between viable candidate words in the cohort.

Empirical evidence against the FUL model arose among others from studies on assimilation (Gaskell & Marslen-Wilson, 1996; Gow, 2001). In a study very similar to the one by Wheeldon and Waksler (2004), Gaskell and Marslen-Wilson (1996) embedded an auditory prime word in a sentence, followed by a visual target word for lexical decision. The prime word either ended correctly with a coronal consonant (e.g. ‘We have a house full of fussy eaters. Sandra will only eat LEAN bacon.’), or the final consonant was changed into a non-coronal one (e.g.

‘.... Sandra will only eat LEAM bacon.’). The control condition had a completely different prime word (e.g. ‘.... Sandra will only eat BROWN loaves.’). Immediately after the prime word was played, the visual target (e.g. LEAN, in all three conditions) appeared on the screen. In the first experiment, the sentences were only presented up to (and including) the prime word. Reaction time results revealed equally speeded responses in the unchanged and changed conditions as compared to the control condition. In the second experiment, the sentences were presented in full length, and the word following the prime word either licensed the change of the last consonant in terms of assimilation (e.g. ‘.... Sandra will only eat LEAM bacon.’), or did not license the change (e.g. ‘.... Sandra will only eat LEAM gammon.’). This time reaction times revealed equally facilitated responses in the unchanged condition and the changed condition with a viable context, but no facilitation in the changed condition in the inappropriate context. This clearly contradicts the findings of Wheeldon and Waksler (2004), who did not find any effects of context. Gaskell and Marslen-Wilson (1996) explained these results in terms of a conservative matching process, which only punishes an apparent mismatch in the light of clear evidence. In the first experiment with the incomplete

sentences, it was unclear whether the changed leam was a mismatch to the coronal-final word lean in the lexicon, or whether this change was due to assimilatory processes. In the second experiment, the inappropriate context (e.g. ‘.... Sandra will only eat LEAM gammon.’) did not license the assumption that /m/ in leam was caused by assimilation of lean, and hence the processor did not accept leam as a variant of lean. On the contrary, the FUL model assumes that the feature [CORONAL] is unspecified and no mismatch ever occurs.

Gow (2001) went one step further and investigated phonologically implausible primes, i.e. he changed the final non-coronal consonant of a prime word into a coronal one, a process that does not usually happen in assimilation. For instance, mispronounced bloom in the sentence ‘The first bloon dies off with a late frost’ speeded reaction times for the visual target word BLOOM as compared to the unrelated target word CLAIM. (Note that one usually varies the prime word rather than the target word between experimental and control condition in order to calculate response times for the same words across groups). The FUL model assumes that the feature [LABIAL] in the /m/ of bloom is specified in the mental lexicon and hence bloon with coronal place information in the signal should not prime the underlying form bloom. Moreover, reaction times did not differ between this phonologically implausible condition and the phonologically plausible condition in which a word-final coronal consonant (e.g. green) was replaced by a labial or dorsal consonant (e.g. greem bag).

The experimental setup of Gow’s (2001) experiment again closely resembles the one used by Wheeldon and Waksler (2004). While they found significant differences in reaction time between a nomismatch and a mismatch in their change-condition, Gow reports equal priming effects for both. In both studies the prime word was embedded in a sentence context. However, in at least some of the sentences used by Gow, the context prior to the prime word allowed expectations with regard to its nature (e.g.: ‘The coverage of the terrible crime drew the wrath of the critics.’ ‘To stay warm they had to burn poles (…)’. Gow, 2001, p.156). This could have activated the target word independently of the phonological input and consequently blurred the response time patterns.

In most priming studies reported so far (Gaskell & Marslen-Wilson, 1996;

Lahiri & van Coillie, 1999; Gow, 2001; Wheeldon & Waksler, 2004; Felder, 2006;

Friedrich, Lahiri & Eulitz, 2008), the prime word always turned into a pseudoword when a segment was changed for place of articulation (e.g. bloom - *bloon).

Consequently, no lexical competition has taken place between the changed prime and the target word. Gow (2002) conducted a further priming experiment in which

the change of the word-final consonant turned this prime-word into another, already existing word. His intention was to test whether coronal place assimilation creates lexical ambiguity as the FUL model would predict. He embedded the prime items in sentence contexts that could induce labialisation of the final segment (e.g.

‘This time she tried to get the right berries for her pie.’) After hearing the prime word in unchanged (e.g. right) or assimilated (e.g. ripe) form, a target word appeared on the screen. This was either the original form (e.g. RIGHT), the equally possible assimilated form (e.g. RIPE) or an unrelated item (e.g. HEM or HEN).

Participants hearing the assimilated token of the word right that perceptually approximated ripe responded faster to the target word RIGHT, but not RIPE. That means, listeners had accessed only the intended word. The FUL model predicted that ripe activates both forms, right and ripe.

According to Gow (2002), assimilation produces a range of modifications, rather than discrete feature changes. That is, assimilated items show a pattern of formant movement that is intermediate between the patterns associated with unmodified coronal or non-coronal place of articulation. Therefore, with sufficient information about coronality left in the signal, no lexical ambiguity would arise in his experiment and listeners know that the intended word was coronal-final right rather than labial-final ripe. However, if this is indeed the case, also the FUL model could explain these results. Leaving the listener with sufficient information about the nonassimilated value of the place feature (i.e. [CORONAL]), implies that coronality is indeed extracted from the signal and hence will mismatch with the labial specification of the /p/ in ripe.

Gaskell and Marslen-Wilson (2001) conducted a similar experiment as the one described above (Gaskell & Marslen-Wilson, 1996), using real words only.

They presented sentences with prime words ending in coronal or non-coronal consonants, and with a following word that either allowed for assimilation or did not. For instance, in a sentence like ‘I think a quick rum picks you up’, rum could be interpreted either as rum, or as a place assimilated form of run. On the other hand, the context in ‘I think a quick rum does you good’ favours the interpretation of rum as a non-assimilated form, since no labial consonant follows. Response times to the visual target (RUM or RUN) presented after the prime word showed that only the respective surface form was activated (i.e. RUN after a sentence like I think a quick run picks you up, and RUM after a sentence like I think a quick rum picks you up). There was no effect of the following word and the possibility of assimilation. This contradicts the findings of Gaskell and Marslen-Wilson (1996) and suggests that the conservative mapping process only applies if there is no word

that matches the surface form directly. However, if the critical sentence was preceded by another sentence that biased the interpretation towards the coronal-final target (e.g. It’s best to start the day with a burst of activity. I think a quick rum picks you up.), then responses to the target RUN were facilitated, suggesting that assimilation was taken into account and run was accessed in the mental lexicon.

Obviously, a lot of factors in the experimental designs seem to influence the processes of lexical activation, or at least the outcome of the experiments pro and contra the FUL model. Among others, the effects of semantic and segmental context, of whether the prime item is a fragment, a pseudoword or an existing word, the effects of competitor size, of vowel vs. consonant change, of phonetic detail in the acoustic signal etc. should be considered. As a first step, in the following experiment, we again used the cross modal fragment priming design as did Friedrich and colleagues (Friedrich, Lahiri & Eulitz, 2008) and altered two factors: We investigated the underspecification of the place feature [CORONAL] in vowels instead of consonants and used prime fragments that activated existing German words, rather than pseudowords.

3.5 Experiment 3: EEG Study on Underspecified Lexical