TL;DR: Science has two syllables compared with just one in signs, but phonologic factors like fast-speech rules and characteristics of Southeast Asian languages might make them sound alike you.
When you ask “how many syllables” a word has, especially one like science, you open up an extremely broad question whose complete treatment is probably beyond the scope of the SE format. This, therefore, is merely an abbreviated outline of the topic.
Phonemically, the word science is /ˈsaɪəns/ in English, while the word signs is /ˈsaɪnz/. Because these two phonemic sequences differ slightly, these are not homophones in the minds of native speakers. However, in actual practice the sounds produced can in certain contexts be virtually indistinguishable to certain listeners. This seems to be what is happening to you.
The mapping of abstract phonemes into phonetic realizations — that is, the mapping of something that exists only in one’s mind into physical sounds — and back again can vary dramatically according to many different factors, including local accent characteristics and what kind of speech the words occur in. All of this probably factors into your own apprehension; it may also contribute to the actual sound production as well.
Asking someone to say a word in isolation will always produce results that are quite different from those produced by the same word spoken as part of a larger phrase.
- Part of this is due to exaggeration of phonetic features that the speaker feels important to make apparent in the careful articulation of an isolated instance of a word.
- Another part of this is that a word that is part of a larger utterance now has other rules shaping it, including rules of phrasal prosody and phonologic phenomena due to the words around it.
If you ask a native speaker what the two syllables are in science, they will tell you that the first one is [ˈsaɪ] and that the second one is something like [ˈɛns], [ˈɪns], or [ˈʌns].
Notice that something strange has happened to the second syllable: the /ə/ cannot become [ə], because when broken apart and utterly separately, the two syllables each become isolated monosyllables. One-word utterances are always stressed, but [ə] can occur only in unstressed syllables in English, never in stressed ones. You can look at [ʌ] as the stressed version of unstressed [ə].
Those who say that the second syllable is [ˈɛns] would be using the so-called spelling pronunciation for that syllable. The problem is that it isn’t actually ever said that way. It might be reduced to [ɪ] or further to [ɨ], but an unstressed [ɛ] will not survive reduction intact. If there were another word with the e spelled with an a, so *sciance, this word would be pronounced just like science is. That’s why dependent and dependant have the same pronunciation.
On the other hand, if you heard the word in casual speech, particularly in rapid speech, then entirely different phonetic rules apply, and you may well be mismapping the resulting sounds back into logical phonemic sequences in your own head.
Because English is a stress-timed language, it is subject to many smaller rules that may affect your perception of the sounds, especially under fast speech rules, also known as allegro rules in linguistics literature. As Professor Lawler mentions, some of the rules that apply to fast speech in English include assimilation, cluster simplification, unstressed vowel reduction, and indeed outright deletion of unstressed syllables. All these will affect how the word science is pronounced in real speech.
However, there may be another factor at work here that just fast-speech rules alone. You say that you are from Southeast Asia. Per the Wikipedia article on the Mainland Southeast Asia linguistic area, languages from that area are especially notable for their syllabic and tonal characteristics:
Syllable structure: A characteristic of MSEA languages is a particular syllable structure involving monosyllabic morphemes, lexical tone, a fairly large inventory of consonants, including phonemic aspiration, limited clusters at the beginning of a syllable, and plentiful vowel contrasts. Final consonants are typically highly restricted, often limited to glides and nasals or unreleased stops at the same points of articulation, with no clusters and no voice distinction. Languages in the northern part of the area generally have fewer vowel and final contrasts but more initial contrasts. [. . .]
Tone Systems: Phonemic tone is one of the most well-known of southeast Asian language characteristics. The tone systems of Middle Chinese, Proto-Miao–Yao, proto-Tai and early Vietnamese all display a three-way tonal contrast in syllables lacking stop endings. In traditional analyses, syllables ending in stops have been treated as a fourth or "checked tone", because their distribution parallels that of syllables with nasal codas.
These two characteristics interact to alter how native speakers of those languages perceive syllables, which in turn alters how they produce them and how listeners in turn map those sounds back into abstract structures in their own minds. Those differences that occur between science and signs, including what happens to them in fast speech, are phonologic aspects that are particularly relevant to languages in the MSEA area.
There are no stops — no obstruents — at all in either of science or signs. The two syllables of science are a stressed open syllable with a diphthong followed by an unstressed closed syllable with a nasal cluster in the coda. There is therefore no consonant boundary between those two syllables, unless one forces one by thinking of the /aɪ/ diphthong as an /aj/ sequence so that you have a semi-consonantal glide between the stressed a of the first syllable and the unstressed /ə/ of the second.
When spoken in a larger utterance, especially quickly, the phonemically contrastive difference in voicing between unvoiced /s/ and voiced /z/ can be erased due to regressive assimilation.
In other words, the one that is normally voiced at the end of one word may become unvoiced depending on how the word that follows it begins, while the one that is normally unvoiced at the end may become voiced because of the start of the next word in the phrase.
In rapid speech, that unstressed /əns/ syllable can further reduce. You might lose the /n/ component and be left with [ə̃s] or even just [əs]. You will almost certainly seem to lose the [ə]. It may be that you will be left with a syllabic consonant producing [n̩], but it may mean that that segment’s syllabic nucleus just disappeared altogether. Which means you just lost a syllable.
Just as different phonetic rules apply when words are used in carefully enunciated isolation compared to those that apply in fast speech, another set of rules applies to metered, rhymed verse: those of metrical phonology. This can be useful for demonstrating the rime of pairs of words in ways that can counter, or at least reduce, certain local accent rules and other matters of pitch and accentuation, much as song can do.
You can force a two-syllable rhyming interpretation using the following bit of doggerel in trochaic tetrameter:
If you place your trust in science,
Might this be too much reliance
On the written words of giants?
Because it’s trochaic, you have four equal feet each with two syllables: STRONG–weak.
When forced into this meter, all of in science, reliance, and of giants rhyme. Moreover, the meter forces a two-syllable interpretation of science and giants even in accents that would normally use one-syllable versions of those, such as those characteristic of the Deep South of the United States.
